v0.13.3

When using native tool calling, the assistant's tool_calls must be saved in
message history. Without this, the model doesn't know it already called a
tool and may loop indefinitely calling the same tool.

This fix adds tool_calls forwarding in chatStreamResponse and chatWithOptions
history formatting.

.serena/.gitignore

@@ -1 +0,0 @@
-/cache

.serena/project.yml

@@ -1,67 +0,0 @@
-# language of the project (csharp, python, rust, java, typescript, go, cpp, or ruby)
-#  * For C, use cpp
-#  * For JavaScript, use typescript
-# Special requirements:
-#  * csharp: Requires the presence of a .sln file in the project folder.
-language: typescript
-
-# whether to use the project's gitignore file to ignore files
-# Added on 2025-04-07
-ignore_all_files_in_gitignore: true
-# list of additional paths to ignore
-# same syntax as gitignore, so you can use * and **
-# Was previously called `ignored_dirs`, please update your config if you are using that.
-# Added (renamed) on 2025-04-07
-ignored_paths: []
-
-# whether the project is in read-only mode
-# If set to true, all editing tools will be disabled and attempts to use them will result in an error
-# Added on 2025-04-18
-read_only: false
-
-# list of tool names to exclude. We recommend not excluding any tools, see the readme for more details.
-# Below is the complete list of tools for convenience.
-# To make sure you have the latest list of tools, and to view their descriptions, 
-# execute `uv run scripts/print_tool_overview.py`.
-#
-#  * `activate_project`: Activates a project by name.
-#  * `check_onboarding_performed`: Checks whether project onboarding was already performed.
-#  * `create_text_file`: Creates/overwrites a file in the project directory.
-#  * `delete_lines`: Deletes a range of lines within a file.
-#  * `delete_memory`: Deletes a memory from Serena's project-specific memory store.
-#  * `execute_shell_command`: Executes a shell command.
-#  * `find_referencing_code_snippets`: Finds code snippets in which the symbol at the given location is referenced.
-#  * `find_referencing_symbols`: Finds symbols that reference the symbol at the given location (optionally filtered by type).
-#  * `find_symbol`: Performs a global (or local) search for symbols with/containing a given name/substring (optionally filtered by type).
-#  * `get_current_config`: Prints the current configuration of the agent, including the active and available projects, tools, contexts, and modes.
-#  * `get_symbols_overview`: Gets an overview of the top-level symbols defined in a given file.
-#  * `initial_instructions`: Gets the initial instructions for the current project.
-#     Should only be used in settings where the system prompt cannot be set,
-#     e.g. in clients you have no control over, like Claude Desktop.
-#  * `insert_after_symbol`: Inserts content after the end of the definition of a given symbol.
-#  * `insert_at_line`: Inserts content at a given line in a file.
-#  * `insert_before_symbol`: Inserts content before the beginning of the definition of a given symbol.
-#  * `list_dir`: Lists files and directories in the given directory (optionally with recursion).
-#  * `list_memories`: Lists memories in Serena's project-specific memory store.
-#  * `onboarding`: Performs onboarding (identifying the project structure and essential tasks, e.g. for testing or building).
-#  * `prepare_for_new_conversation`: Provides instructions for preparing for a new conversation (in order to continue with the necessary context).
-#  * `read_file`: Reads a file within the project directory.
-#  * `read_memory`: Reads the memory with the given name from Serena's project-specific memory store.
-#  * `remove_project`: Removes a project from the Serena configuration.
-#  * `replace_lines`: Replaces a range of lines within a file with new content.
-#  * `replace_symbol_body`: Replaces the full definition of a symbol.
-#  * `restart_language_server`: Restarts the language server, may be necessary when edits not through Serena happen.
-#  * `search_for_pattern`: Performs a search for a pattern in the project.
-#  * `summarize_changes`: Provides instructions for summarizing the changes made to the codebase.
-#  * `switch_modes`: Activates modes by providing a list of their names
-#  * `think_about_collected_information`: Thinking tool for pondering the completeness of collected information.
-#  * `think_about_task_adherence`: Thinking tool for determining whether the agent is still on track with the current task.
-#  * `think_about_whether_you_are_done`: Thinking tool for determining whether the task is truly completed.
-#  * `write_memory`: Writes a named memory (for future reference) to Serena's project-specific memory store.
-excluded_tools: []
-
-# initial prompt for the project. It will always be given to the LLM upon activating the project
-# (contrary to the memories, which are loaded on demand).
-initial_prompt: ""
-
-project_name: "smartai"

changelog.md

@@ -1,5 +1,68 @@
 # Changelog
 
+## 2026-01-20 - 0.13.3 - fix()
+no changes detected
+
+- No files changed in the provided diff.
+- No version bump required.
+
+## 2026-01-20 - 0.13.2 - fix(repo)
+no changes detected in diff; nothing to commit
+
+- Git diff reported no changes — no files modified
+- No code or dependency updates detected, so no version bump required
+
+## 2026-01-20 - 0.13.1 - fix()
+no changes detected; no release required
+
+- No changes found in the provided git diff
+- Current package version is 0.13.0
+
+## 2026-01-20 - 0.13.0 - feat(provider.ollama)
+add chain-of-thought reasoning support to chat messages and Ollama provider
+
+- Added optional reasoning?: string to chat message and chat response interfaces to surface chain-of-thought data.
+- Propagates reasoning from message history into formatted requests sent to Ollama.
+- Maps Ollama response fields (thinking or reasoning) into ChatResponse.reasoning so downstream code can access model reasoning output.
+
+## 2026-01-20 - 0.12.1 - fix(docs)
+update documentation: clarify provider capabilities, add provider capabilities summary, polish examples and formatting, and remove Serena project config
+
+- Removed .serena/project.yml and cleaned up .serena/.gitignore
+- Added Provider Capabilities Summary and expanded/clarified provider tables in readme.md and readme.hints.md
+- Clarified Anthropic extended thinking details and Mistral native PDF OCR notes
+- Polished example code snippets and fixed minor typos/formatting (GPT-5 mention, ElevenLabs model note, consistent punctuation)
+- Updated test command references and other README usage instructions
+
+## 2026-01-20 - 0.12.0 - feat(ollama)
+add support for base64-encoded images in chat messages and forward them to the Ollama provider
+
+- Add optional images?: string[] to ChatMessage and ChatOptions interfaces (multimodal/vision support)
+- Propagate images from messageHistory and ChatOptions to the Ollama API payload in chat, chatStreaming, and streaming handlers
+- Changes are non-breaking: images are optional and existing behavior is preserved when absent
+
+## 2026-01-20 - 0.11.0 - feat(ollama)
+support defaultOptions and defaultTimeout for ollama provider
+
+- Added ollama.defaultOptions object with fields: num_ctx, temperature, top_k, top_p, repeat_penalty, num_predict, stop, seed
+- Added ollama.defaultTimeout option
+- Pass defaultOptions and defaultTimeout into OllamaProvider constructor when initializing the provider
+- Non-breaking change: existing behavior preserved if new fields are undefined
+
+## 2026-01-20 - 0.10.1 - fix()
+no changes detected — no release necessary
+
+- No files changed in the provided diff; there are no code, documentation, or configuration modifications to release.
+
+## 2026-01-18 - 0.10.0 - feat(mistral)
+add Mistral provider with native PDF OCR and chat integration
+
+- Adds dependency @mistralai/mistralai
+- Implements ts/provider.mistral.ts providing chat() and document() (OCR) functionality
+- Registers and exposes MistralProvider in SmartAi (options, lifecycle, conversation routing)
+- Adds unit/integration tests: test.chat.mistral.ts and test.document.mistral.ts
+- Updates readme.hints.md with Mistral usage, configuration and notes
+
 ## 2026-01-18 - 0.9.0 - feat(providers)
 Add Anthropic extended thinking and adapt providers to new streaming/file APIs; bump dependencies and update docs, tests and configuration
 

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@push.rocks/smartai",
-  "version": "0.9.0",
+  "version": "0.13.3",
   "private": false,
   "description": "SmartAi is a versatile TypeScript library designed to facilitate integration and interaction with various AI models, offering functionalities for chat, audio generation, document processing, and vision tasks.",
   "main": "dist_ts/index.js",
@@ -18,21 +18,22 @@
     "@git.zone/tsbuild": "^4.1.2",
     "@git.zone/tsbundle": "^2.8.1",
     "@git.zone/tsrun": "^2.0.1",
-    "@git.zone/tstest": "^3.1.4",
+    "@git.zone/tstest": "^3.1.6",
     "@push.rocks/qenv": "^6.1.3",
-    "@types/node": "^22.15.17",
+    "@types/node": "^25.0.9",
     "typescript": "^5.9.3"
   },
   "dependencies": {
     "@anthropic-ai/sdk": "^0.71.2",
+    "@mistralai/mistralai": "^1.12.0",
     "@push.rocks/smartarray": "^1.1.0",
-    "@push.rocks/smartfile": "^11.2.7",
+    "@push.rocks/smartfs": "^1.3.1",
     "@push.rocks/smartpath": "^6.0.0",
     "@push.rocks/smartpdf": "^4.1.1",
     "@push.rocks/smartpromise": "^4.2.3",
     "@push.rocks/smartrequest": "^5.0.1",
     "@push.rocks/webstream": "^1.0.10",
-    "openai": "^5.12.2"
+    "openai": "^6.16.0"
   },
   "repository": {
     "type": "git",

pnpm-lock.yaml

@@ -11,12 +11,15 @@ importers:
       '@anthropic-ai/sdk':
         specifier: ^0.71.2
         version: 0.71.2(zod@3.25.76)
+      '@mistralai/mistralai':
+        specifier: ^1.12.0
+        version: 1.12.0
       '@push.rocks/smartarray':
         specifier: ^1.1.0
         version: 1.1.0
-      '@push.rocks/smartfile':
+      '@push.rocks/smartfs':
-        specifier: ^11.2.7
+        specifier: ^1.3.1
-        version: 11.2.7
+        version: 1.3.1
       '@push.rocks/smartpath':
         specifier: ^6.0.0
         version: 6.0.0
@@ -33,8 +36,8 @@ importers:
         specifier: ^1.0.10
         version: 1.0.10
       openai:
-        specifier: ^5.12.2
+        specifier: ^6.16.0
-        version: 5.12.2(ws@8.18.3)(zod@3.25.76)
+        version: 6.16.0(ws@8.18.3)(zod@3.25.76)
     devDependencies:
       '@git.zone/tsbuild':
         specifier: ^4.1.2
@@ -46,14 +49,14 @@ importers:
         specifier: ^2.0.1
         version: 2.0.1
       '@git.zone/tstest':
-        specifier: ^3.1.4
+        specifier: ^3.1.6
-        version: 3.1.4(@aws-sdk/credential-providers@3.808.0)(socks@2.8.4)(typescript@5.9.3)
+        version: 3.1.6(@aws-sdk/credential-providers@3.808.0)(socks@2.8.4)(typescript@5.9.3)
       '@push.rocks/qenv':
         specifier: ^6.1.3
         version: 6.1.3
       '@types/node':
-        specifier: ^22.15.17
+        specifier: ^25.0.9
-        version: 22.15.17
+        version: 25.0.9
       typescript:
         specifier: ^5.9.3
         version: 5.9.3
@@ -813,8 +816,8 @@ packages:
     resolution: {integrity: sha512-NEcnsjvlC1o3Z6SS3VhKCf6Ev+Sh4EAinmggslrIR/ppMrvjDbXNFXoyr3PB+GLeSAR0JRZ1fGvVYjpEzjBdIg==}
     hasBin: true
 
-  '@git.zone/tstest@3.1.4':
+  '@git.zone/tstest@3.1.6':
-    resolution: {integrity: sha512-S7kubbb0yLYOh/QAzFsjG6a20lZiyNKo4pt0yK1yvd9I7X8Rw6/mCT/BicLkan7G7Nk7scUfxaK9+aFsHmdQdw==}
+    resolution: {integrity: sha512-xRGc6wO4rJ6mohPCMIBDRH+oNjiIvX6Jeo8v/Y5o5VyKSHFmqol7FCKSBrojMcqgBpESnLHFPJAAOmT9W3JV8Q==}
     hasBin: true
 
   '@happy-dom/global-registrator@15.11.7':
@@ -1132,6 +1135,9 @@ packages:
   '@lit/reactive-element@2.1.1':
     resolution: {integrity: sha512-N+dm5PAYdQ8e6UlywyyrgI2t++wFGXfHx+dSJ1oBrg6FAxUj40jId++EaRm80MKX5JnlH1sBsyZ5h0bcZKemCg==}
 
+  '@mistralai/mistralai@1.12.0':
+    resolution: {integrity: sha512-oDr1hcS3wsIT/QupBG93TNiA5kilwBYoAIyl5BNYqMM2Ix/xsNq+wT8b++uhp/GTUMx44n+8Bn1mkATbwxe6bQ==}
+
   '@mixmark-io/domino@2.2.0':
     resolution: {integrity: sha512-Y28PR25bHXUg88kCV7nivXrP2Nj2RueZ3/l/jdx6J9f8J4nsEGcgX0Qe6lt7Pa+J79+kPiJU3LguR6O/6zrLOw==}
 
@@ -2491,8 +2497,11 @@ packages:
   '@types/node@16.9.1':
     resolution: {integrity: sha512-QpLcX9ZSsq3YYUUnD3nFDY8H7wctAhQj/TFKL8Ya8v5fMm3CFXxo8zStsLAl780ltoYoo1WvKUVGBQK+1ifr7g==}
 
-  '@types/node@22.15.17':
+  '@types/node@22.19.7':
-    resolution: {integrity: sha512-wIX2aSZL5FE+MR0JlvF87BNVrtFWf6AE6rxSE9X7OwnVvoyCQjpzSRJ+M87se/4QCkCiebQAqrJ0y6fwIyi7nw==}
+    resolution: {integrity: sha512-MciR4AKGHWl7xwxkBa6xUGxQJ4VBOmPTF7sL+iGzuahOFaO0jHCsuEfS80pan1ef4gWId1oWOweIhrDEYLuaOw==}
+
+  '@types/node@25.0.9':
+    resolution: {integrity: sha512-/rpCXHlCWeqClNBwUhDcusJxXYDjZTyE8v5oTO7WbL8eij2nKhUeU89/6xgjU7N4/Vh3He0BtyhJdQbDyhiXAw==}
 
   '@types/ping@0.4.4':
     resolution: {integrity: sha512-ifvo6w2f5eJYlXm+HiVx67iJe8WZp87sfa683nlqED5Vnt9Z93onkokNoWqOG21EaE8fMxyKPobE+mkPEyxsdw==}
@@ -4019,12 +4028,12 @@ packages:
     resolution: {integrity: sha512-7x81NCL719oNbsq/3mh+hVrAWmFuEYUqrq/Iw3kUzH8ReypT9QQ0BLoJS7/G9k6N81XjW4qHWtjWwe/9eLy1EQ==}
     engines: {node: '>=12'}
 
-  openai@5.12.2:
+  openai@6.16.0:
-    resolution: {integrity: sha512-xqzHHQch5Tws5PcKR2xsZGX9xtch+JQFz5zb14dGqlshmmDAFBFEWmeIpf7wVqWV+w7Emj7jRgkNJakyKE0tYQ==}
+    resolution: {integrity: sha512-fZ1uBqjFUjXzbGc35fFtYKEOxd20kd9fDpFeqWtsOZWiubY8CZ1NAlXHW3iathaFvqmNtCWMIsosCuyeI7Joxg==}
     hasBin: true
     peerDependencies:
       ws: ^8.18.0
-      zod: ^3.23.8
+      zod: ^3.25 || ^4.0
     peerDependenciesMeta:
       ws:
         optional: true
@@ -4715,6 +4724,9 @@ packages:
   undici-types@6.21.0:
     resolution: {integrity: sha512-iwDZqg0QAGrg9Rav5H4n0M64c3mkR59cJ6wQp+7C4nI0gsmExaedaYLNO44eT4AtBBwjbTiGPMlt2Md0T9H9JQ==}
 
+  undici-types@7.16.0:
+    resolution: {integrity: sha512-Zz+aZWSj8LE6zoxD+xrjh4VfkIG8Ya6LvYkZqtUQGJPZjYl53ypCaUwWqo7eI0x66KBGeRo+mlBEkMSeSZ38Nw==}
+
   unified@11.0.5:
     resolution: {integrity: sha512-xKvGhPWw3k84Qjh8bI3ZeJjqnyadK+GEFtazSfZv/rKeTkTjOJho6mFqh2SM96iIcZokxiOpg78GazTSg8+KHA==}
 
@@ -4895,6 +4907,11 @@ packages:
     resolution: {integrity: sha512-U/PBtDf35ff0D8X8D0jfdzHYEPFxAI7jJlxZXwCSez5M3190m+QobIfh+sWDWSHMCWWJN2AWamkegn6vr6YBTw==}
     engines: {node: '>=18'}
 
+  zod-to-json-schema@3.25.1:
+    resolution: {integrity: sha512-pM/SU9d3YAggzi6MtR4h7ruuQlqKtad8e9S0fmxcMi+ueAK5Korys/aWcV9LIIHTVbj01NdzxcnXSN+O74ZIVA==}
+    peerDependencies:
+      zod: ^3.25 || ^4
+
   zod@3.24.2:
     resolution: {integrity: sha512-lY7CDW43ECgW9u1TcT3IoXHflywfVqDYze4waEz812jR/bZ8FHDsl7pFQoSZTz5N+2NqRXs8GBwnAwo3ZNxqhQ==}
 
@@ -6906,7 +6923,7 @@ snapshots:
       '@push.rocks/smartshell': 3.3.0
       tsx: 4.21.0
 
-  '@git.zone/tstest@3.1.4(@aws-sdk/credential-providers@3.808.0)(socks@2.8.4)(typescript@5.9.3)':
+  '@git.zone/tstest@3.1.6(@aws-sdk/credential-providers@3.808.0)(socks@2.8.4)(typescript@5.9.3)':
     dependencies:
       '@api.global/typedserver': 3.0.80
       '@git.zone/tsbundle': 2.8.1
@@ -7060,7 +7077,7 @@ snapshots:
       '@inquirer/figures': 1.0.15
       '@inquirer/type': 2.0.0
       '@types/mute-stream': 0.0.4
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
       '@types/wrap-ansi': 3.0.0
       ansi-escapes: 4.3.2
       cli-width: 4.1.0
@@ -7350,6 +7367,11 @@ snapshots:
     dependencies:
       '@lit-labs/ssr-dom-shim': 1.4.0
 
+  '@mistralai/mistralai@1.12.0':
+    dependencies:
+      zod: 3.25.76
+      zod-to-json-schema: 3.25.1(zod@3.25.76)
+
   '@mixmark-io/domino@2.2.0': {}
 
   '@module-federation/error-codes@0.22.0': {}
@@ -9635,27 +9657,27 @@ snapshots:
 
   '@types/bn.js@5.2.0':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/body-parser@1.19.6':
     dependencies:
       '@types/connect': 3.4.38
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/buffer-json@2.0.3': {}
 
   '@types/clean-css@4.2.11':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
       source-map: 0.6.1
 
   '@types/connect@3.4.38':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/cors@2.8.19':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/debug@4.1.12':
     dependencies:
@@ -9665,7 +9687,7 @@ snapshots:
 
   '@types/dns-packet@5.6.5':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/elliptic@6.4.18':
     dependencies:
@@ -9673,7 +9695,7 @@ snapshots:
 
   '@types/express-serve-static-core@5.0.7':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
       '@types/qs': 6.14.0
       '@types/range-parser': 1.2.7
       '@types/send': 0.17.5
@@ -9697,16 +9719,16 @@ snapshots:
   '@types/fs-extra@11.0.4':
     dependencies:
       '@types/jsonfile': 6.1.4
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/fs-extra@9.0.13':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/glob@7.2.0':
     dependencies:
       '@types/minimatch': 5.1.2
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/hast@3.0.4':
     dependencies:
@@ -9728,7 +9750,7 @@ snapshots:
 
   '@types/jsonfile@6.1.4':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/mdast@4.0.4':
     dependencies:
@@ -9746,18 +9768,22 @@ snapshots:
 
   '@types/mute-stream@0.0.4':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/node-forge@1.3.11':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/node@16.9.1': {}
 
-  '@types/node@22.15.17':
+  '@types/node@22.19.7':
     dependencies:
       undici-types: 6.21.0
 
+  '@types/node@25.0.9':
+    dependencies:
+      undici-types: 7.16.0
+
   '@types/ping@0.4.4': {}
 
   '@types/qs@6.14.0': {}
@@ -9773,28 +9799,28 @@ snapshots:
   '@types/send@0.17.5':
     dependencies:
       '@types/mime': 1.3.5
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/serve-static@1.15.8':
     dependencies:
       '@types/http-errors': 2.0.5
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
       '@types/send': 0.17.5
 
   '@types/serve-static@2.2.0':
     dependencies:
       '@types/http-errors': 2.0.5
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/symbol-tree@3.2.5': {}
 
   '@types/tar-stream@3.1.4':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/through2@2.0.41':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/trusted-types@2.0.7': {}
 
@@ -9816,7 +9842,7 @@ snapshots:
 
   '@types/whatwg-url@8.2.2':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
       '@types/webidl-conversions': 7.0.3
 
   '@types/which@3.0.4': {}
@@ -9825,11 +9851,11 @@ snapshots:
 
   '@types/ws@8.18.1':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
 
   '@types/yauzl@2.10.3':
     dependencies:
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
     optional: true
 
   '@ungap/structured-clone@1.3.0': {}
@@ -10289,7 +10315,7 @@ snapshots:
   engine.io@6.6.4:
     dependencies:
       '@types/cors': 2.8.19
-      '@types/node': 22.15.17
+      '@types/node': 22.19.7
       accepts: 1.3.8
       base64id: 2.0.0
       cookie: 0.7.2
@@ -11556,7 +11582,7 @@ snapshots:
       is-docker: 2.2.1
       is-wsl: 2.2.0
 
-  openai@5.12.2(ws@8.18.3)(zod@3.25.76):
+  openai@6.16.0(ws@8.18.3)(zod@3.25.76):
     optionalDependencies:
       ws: 8.18.3
       zod: 3.25.76
@@ -12351,6 +12377,8 @@ snapshots:
 
   undici-types@6.21.0: {}
 
+  undici-types@7.16.0: {}
+
   unified@11.0.5:
     dependencies:
       '@types/unist': 3.0.3
@@ -12506,6 +12534,10 @@ snapshots:
 
   yoctocolors-cjs@2.1.3: {}
 
+  zod-to-json-schema@3.25.1(zod@3.25.76):
+    dependencies:
+      zod: 3.25.76
+
   zod@3.24.2: {}
 
   zod@3.25.76: {}

readme.hints.md

@@ -3,8 +3,10 @@
 ## Dependencies
 
 - Uses `@git.zone/tstest` v3.x for testing (import from `@git.zone/tstest/tapbundle`)
-- `@push.rocks/smartfile` is kept at v11 to avoid migration to factory pattern
+- `@push.rocks/smartfs` v1.x for file system operations
 - `@anthropic-ai/sdk` v0.71.x with extended thinking support
+- `@mistralai/mistralai` v1.x for Mistral OCR and chat capabilities
+- `openai` v6.x for OpenAI API integration
 - `@push.rocks/smartrequest` v5.x - uses `response.stream()` + `Readable.fromWeb()` for streaming
 
 ## Important Notes
@@ -12,11 +14,52 @@
 - When extended thinking is enabled, temperature parameter must NOT be set (or set to 1)
 - The `streamNode()` method was removed in smartrequest v5, use `response.stream()` with `Readable.fromWeb()` instead
 
-## Anthropic Extended Thinking Feature
+## Provider Capabilities Summary
+
+| Provider     | Chat | Stream | TTS | Vision | Documents | Research | Images |
+|--------------|------|--------|-----|--------|-----------|----------|--------|
+| OpenAI       | ✅   | ✅     | ✅  | ✅     | ✅        | ✅       | ✅     |
+| Anthropic    | ✅   | ✅     | ❌  | ✅     | ✅        | ✅       | ❌     |
+| Mistral      | ✅   | ✅     | ❌  | ✅     | ✅        | ❌       | ❌     |
+| ElevenLabs   | ❌   | ❌     | ✅  | ❌     | ❌        | ❌       | ❌     |
+| Ollama       | ✅   | ✅     | ❌  | ✅     | ✅        | ❌       | ❌     |
+| XAI          | ✅   | ✅     | ❌  | ❌     | ✅        | ❌       | ❌     |
+| Perplexity   | ✅   | ✅     | ❌  | ❌     | ❌        | ✅       | ❌     |
+| Groq         | ✅   | ✅     | ❌  | ❌     | ❌        | ❌       | ❌     |
+| Exo          | ✅   | ✅     | ❌  | ❌     | ❌        | ❌       | ❌     |
+
+## Mistral Provider Integration
 
 ### Overview
 
-The Anthropic provider now supports extended thinking by default across all methods. Extended thinking enables Claude to spend more time reasoning about complex problems before generating responses, leading to higher quality answers for difficult questions.
+The Mistral provider supports:
+- **Document AI** via Mistral OCR (December 2025) - native PDF processing without image conversion
+- **Chat capabilities** using Mistral's chat models (`mistral-large-latest`, etc.)
+
+### Key Advantage: Native PDF Support
+
+Unlike other providers that require converting PDFs to images (using SmartPdf), Mistral OCR natively accepts PDF documents as base64-encoded data. This makes document processing potentially faster and more accurate for text extraction.
+
+### Configuration
+
+```typescript
+import * as smartai from '@push.rocks/smartai';
+
+const provider = new smartai.MistralProvider({
+  mistralToken: 'your-token-here',
+  chatModel: 'mistral-large-latest',  // default
+  ocrModel: 'mistral-ocr-latest',     // default
+  tableFormat: 'markdown',             // 'markdown' or 'html'
+});
+
+await provider.start();
+```
+
+### API Key
+
+Tests require `MISTRAL_API_KEY` in `.nogit/env.json`.
+
+## Anthropic Extended Thinking Feature
 
 ### Configuration
 
@@ -33,8 +76,6 @@ const provider = new smartai.AnthropicProvider({
 
 ### Thinking Modes
 
-The `extendedThinking` parameter accepts four modes:
-
 | Mode       | Budget Tokens | Use Case                                        |
 | ---------- | ------------- | ----------------------------------------------- |
 | `'quick'`  | 2,048         | Lightweight reasoning for simple queries        |
@@ -42,141 +83,13 @@ The `extendedThinking` parameter accepts four modes:
 | `'deep'`   | 16,000        | Complex reasoning for difficult problems        |
 | `'off'`    | 0             | Disable extended thinking                       |
 
-**Default Behavior**: If `extendedThinking` is not specified, it defaults to `'normal'` mode (8,000 tokens).
+### Implementation Details
 
-### Supported Methods
+- Extended thinking is implemented via `getThinkingConfig()` private method
+- When thinking is enabled, temperature must NOT be set
+- Uses `claude-sonnet-4-5-20250929` model
 
-Extended thinking is automatically applied to all Anthropic provider methods:
+## Testing
-
-- `chat()` - Synchronous chat
-- `chatStream()` - Streaming chat
-- `vision()` - Image analysis
-- `document()` - PDF document processing
-- `research()` - Web research with citations
-
-### Token Budget Constraints
-
-**Important**: The thinking budget must be less than `max_tokens` for the API call. The current `max_tokens` values are:
-
-- `chatStream()`: 20,000 tokens (sufficient for all modes ✓)
-- `chat()`: 20,000 tokens (sufficient for all modes ✓)
-- `vision()`: 10,000 tokens (sufficient for all modes ✓)
-- `document()`: 20,000 tokens (sufficient for all modes ✓)
-- `research()`: 20,000 tokens for all searchDepth levels (sufficient ✓)
-
-### Performance and Cost Implications
-
-**Token Usage**:
-
-- You are charged for the **full thinking tokens** generated, not just the summary
-- Higher thinking budgets may result in more thorough reasoning but increased costs
-- The budget is a **target**, not a strict limit - actual usage may vary
-
-**Response Quality**:
-
-- `'quick'`: Fast responses, basic reasoning
-- `'normal'`: Good balance between quality and speed (recommended for most use cases)
-- `'deep'`: Highest quality reasoning for complex problems, slower responses
-
-**Recommendations**:
-
-- Start with `'normal'` (default) for general usage
-- Use `'deep'` for complex analytical tasks, philosophy, mathematics, or research
-- Use `'quick'` for simple factual queries where deep reasoning isn't needed
-- Use `'off'` only if you want traditional Claude behavior without extended thinking
-
-### Usage Examples
-
-#### Example 1: Default (Normal Mode)
-
-```typescript
-const provider = new smartai.AnthropicProvider({
-  anthropicToken: process.env.ANTHROPIC_TOKEN,
-  // extendedThinking defaults to 'normal'
-});
-
-await provider.start();
-
-const response = await provider.chat({
-  systemMessage: 'You are a helpful assistant.',
-  userMessage: 'Explain the implications of quantum computing.',
-  messageHistory: [],
-});
-```
-
-#### Example 2: Deep Thinking for Complex Analysis
-
-```typescript
-const provider = new smartai.AnthropicProvider({
-  anthropicToken: process.env.ANTHROPIC_TOKEN,
-  extendedThinking: 'deep', // 16,000 token budget
-});
-
-await provider.start();
-
-const response = await provider.chat({
-  systemMessage: 'You are a philosopher and ethicist.',
-  userMessage: 'Analyze the trolley problem from multiple ethical frameworks.',
-  messageHistory: [],
-});
-```
-
-#### Example 3: Quick Mode for Simple Queries
-
-```typescript
-const provider = new smartai.AnthropicProvider({
-  anthropicToken: process.env.ANTHROPIC_TOKEN,
-  extendedThinking: 'quick', // 2,048 token budget
-});
-
-await provider.start();
-
-const response = await provider.chat({
-  systemMessage: 'You are a helpful assistant.',
-  userMessage: 'What is the capital of France?',
-  messageHistory: [],
-});
-```
-
-#### Example 4: Disable Thinking
-
-```typescript
-const provider = new smartai.AnthropicProvider({
-  anthropicToken: process.env.ANTHROPIC_TOKEN,
-  extendedThinking: 'off', // No extended thinking
-});
-
-await provider.start();
-
-const response = await provider.chat({
-  systemMessage: 'You are a helpful assistant.',
-  userMessage: 'Tell me a joke.',
-  messageHistory: [],
-});
-```
-
-#### Example 5: Extended Thinking with Vision
-
-```typescript
-const provider = new smartai.AnthropicProvider({
-  anthropicToken: process.env.ANTHROPIC_TOKEN,
-  extendedThinking: 'normal',
-});
-
-await provider.start();
-
-const imageBuffer = await fs.promises.readFile('./image.jpg');
-const analysis = await provider.vision({
-  image: imageBuffer,
-  prompt: 'Analyze this image in detail and explain what you see.',
-});
-```
-
-### Testing
-
-Comprehensive tests for extended thinking are available in:
-
-- `test/test.thinking.anthropic.ts` - Tests all thinking modes
 
 Run tests with:
 
@@ -184,27 +97,8 @@ Run tests with:
 pnpm test
 ```
 
-Run specific thinking tests:
+Run specific tests:
 
 ```bash
-npx tstest test/test.thinking.anthropic.ts --verbose
+npx tstest test/test.something.ts --verbose
 ```
-
-### API Reference
-
-According to Anthropic's documentation:
-
-- Extended thinking is supported on Claude Sonnet 4.5, 4, 3.7, Haiku 4.5, and Opus 4.1, 4
-- The current model used is `claude-sonnet-4-5-20250929`
-- Minimum thinking budget is 1,024 tokens
-- Thinking budget must be less than `max_tokens`
-
-### Implementation Details
-
-The extended thinking feature is implemented via:
-
-1. **Interface**: `IAnthropicProviderOptions.extendedThinking` property
-2. **Helper Method**: `getThinkingConfig()` private method that maps modes to token budgets
-3. **API Parameter**: Adds `thinking: { type: 'enabled', budget_tokens: number }` to all API calls
-
-The thinking configuration is applied automatically to all API calls when the provider is instantiated.

readme.md

@@ -6,7 +6,7 @@
 [![TypeScript](https://img.shields.io/badge/TypeScript-5.x-blue.svg)](https://www.typescriptlang.org/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 
-SmartAI unifies the world's leading AI providers - OpenAI, Anthropic, Perplexity, Ollama, Groq, XAI, Exo, and ElevenLabs - under a single, elegant TypeScript interface. Build AI applications at lightning speed without vendor lock-in.
+SmartAI unifies the world's leading AI providers — OpenAI, Anthropic, Mistral, Perplexity, Ollama, Groq, XAI, Exo, and ElevenLabs — under a single, elegant TypeScript interface. Build AI applications at lightning speed without vendor lock-in.
 
 ## Issue Reporting and Security
 
@@ -14,19 +14,23 @@ For reporting bugs, issues, or security vulnerabilities, please visit [community
 
 ## 🎯 Why SmartAI?
 
-- **🔌 Universal Interface**: Write once, run with any AI provider. Switch between GPT-4, Claude, Llama, or Grok with a single line change.
+- **🔌 Universal Interface**: Write once, run with any AI provider. Switch between GPT-5, Claude, Llama, or Grok with a single line change.
-- **🛡️ Type-Safe**: Full TypeScript support with comprehensive type definitions for all operations
+- **🛡️ Type-Safe**: Full TypeScript support with comprehensive type definitions for all operations.
-- **🌊 Streaming First**: Built for real-time applications with native streaming support
+- **🌊 Streaming First**: Built for real-time applications with native streaming support.
-- **🎨 Multi-Modal**: Seamlessly work with text, images, audio, and documents
+- **🎨 Multi-Modal**: Seamlessly work with text, images, audio, and documents.
-- **🏠 Local & Cloud**: Support for both cloud providers and local models via Ollama
+- **🏠 Local & Cloud**: Support for both cloud providers and local models via Ollama/Exo.
-- **⚡ Zero Lock-In**: Your code remains portable across all AI providers
+- **⚡ Zero Lock-In**: Your code remains portable across all AI providers.
 
-## 🚀 Quick Start
+## 📦 Installation
 
 ```bash
 npm install @push.rocks/smartai
+# or
+pnpm install @push.rocks/smartai
 ```
 
+## 🚀 Quick Start
+
 ```typescript
 import { SmartAi } from '@push.rocks/smartai';
 
@@ -48,6 +52,8 @@ const response = await ai.openaiProvider.chat({
   userMessage: 'Explain quantum computing in simple terms',
   messageHistory: [],
 });
+
+console.log(response.message);
 ```
 
 ## 📊 Provider Capabilities Matrix
@@ -56,14 +62,15 @@ Choose the right provider for your use case:
 
 | Provider       | Chat | Streaming | TTS | Vision | Documents | Research | Images | Highlights                                                      |
 | -------------- | :--: | :-------: | :-: | :----: | :-------: | :------: | :----: | --------------------------------------------------------------- |
-| **OpenAI**     |  ✅  |    ✅     | ✅  |   ✅   |    ✅     |    ✅    |   ✅   | • gpt-image-1<br>• DALL-E 3<br>• Deep research API              |
+| **OpenAI**     |  ✅  |    ✅     | ✅  |   ✅   |    ✅     |    ✅    |   ✅   | gpt-image-1 • DALL-E 3 • Deep Research API                      |
-| **Anthropic**  |  ✅  |    ✅     | ❌  |   ✅   |    ✅     |    ✅    |   ❌   | • Claude Sonnet 4.5<br>• Superior reasoning<br>• Web search API |
+| **Anthropic**  |  ✅  |    ✅     | ❌  |   ✅   |    ✅     |    ✅    |   ❌   | Claude Sonnet 4.5 • Extended Thinking • Web Search API          |
-| **ElevenLabs** |  ❌  |    ❌     | ✅  |   ❌   |    ❌     |    ❌    |   ❌   | • Premium TTS<br>• 70+ languages<br>• Natural voices            |
+| **Mistral**    |  ✅  |    ✅     | ❌  |   ✅   |    ✅     |    ❌    |   ❌   | Native PDF OCR • mistral-large • Fast inference                 |
-| **Ollama**     |  ✅  |    ✅     | ❌  |   ✅   |    ✅     |    ❌    |   ❌   | • 100% local<br>• Privacy-first<br>• No API costs               |
+| **ElevenLabs** |  ❌  |    ❌     | ✅  |   ❌   |    ❌     |    ❌    |   ❌   | Premium TTS • 70+ languages • v3 model                          |
-| **XAI**        |  ✅  |    ✅     | ❌  |   ❌   |    ✅     |    ❌    |   ❌   | • Grok models<br>• Real-time data<br>• Uncensored               |
+| **Ollama**     |  ✅  |    ✅     | ❌  |   ✅   |    ✅     |    ❌    |   ❌   | 100% local • Privacy-first • No API costs                       |
-| **Perplexity** |  ✅  |    ✅     | ❌  |   ❌   |    ❌     |    ✅    |   ❌   | • Web-aware<br>• Research-focused<br>• Sonar Pro models         |
+| **XAI**        |  ✅  |    ✅     | ❌  |   ❌   |    ✅     |    ❌    |   ❌   | Grok 2 • Real-time data                                         |
-| **Groq**       |  ✅  |    ✅     | ❌  |   ❌   |    ❌     |    ❌    |   ❌   | • 10x faster<br>• LPU inference<br>• Low latency                |
+| **Perplexity** |  ✅  |    ✅     | ❌  |   ❌   |    ❌     |    ✅    |   ❌   | Web-aware • Research-focused • Sonar Pro                        |
-| **Exo**        |  ✅  |    ✅     | ❌  |   ❌   |    ❌     |    ❌    |   ❌   | • Distributed<br>• P2P compute<br>• Decentralized               |
+| **Groq**       |  ✅  |    ✅     | ❌  |   ❌   |    ❌     |    ❌    |   ❌   | 10x faster • LPU inference • Llama 3.3                          |
+| **Exo**        |  ✅  |    ✅     | ❌  |   ❌   |    ❌     |    ❌    |   ❌   | Distributed • P2P compute • Decentralized                       |
 
 ## 🎮 Core Features
 
@@ -72,9 +79,9 @@ Choose the right provider for your use case:
 Works identically across all providers:
 
 ```typescript
-// Use GPT-4 for complex reasoning
+// Use GPT-5 for complex reasoning
 const gptResponse = await ai.openaiProvider.chat({
-  systemMessage: 'You are a expert physicist.',
+  systemMessage: 'You are an expert physicist.',
   userMessage: 'Explain the implications of quantum entanglement',
   messageHistory: [],
 });
@@ -127,20 +134,17 @@ const audioStream = await ai.openaiProvider.audio({
 const elevenLabsAudio = await ai.elevenlabsProvider.audio({
   message: 'Experience the most lifelike text to speech technology.',
   voiceId: '19STyYD15bswVz51nqLf', // Optional: Samara voice
-  modelId: 'eleven_v3', // Optional: defaults to eleven_v3 (70+ languages, most expressive)
+  modelId: 'eleven_v3', // Optional: defaults to eleven_v3 (70+ languages)
   voiceSettings: {
     // Optional: fine-tune voice characteristics
     stability: 0.5, // 0-1: Speech consistency
     similarity_boost: 0.8, // 0-1: Voice similarity to original
-    style: 0.0, // 0-1: Expressiveness (higher = more expressive)
+    style: 0.0, // 0-1: Expressiveness
     use_speaker_boost: true, // Enhanced clarity
   },
 });
 
-// Stream directly to speakers
+// Stream directly to speakers or save to file
-audioStream.pipe(speakerOutput);
-
-// Or save to file
 audioStream.pipe(fs.createWriteStream('welcome.mp3'));
 ```
 
@@ -157,7 +161,7 @@ const gptVision = await ai.openaiProvider.vision({
   prompt: 'Describe this product and suggest marketing angles',
 });
 
-// Anthropic: Detailed analysis
+// Anthropic: Detailed analysis with extended thinking
 const claudeVision = await ai.anthropicProvider.vision({
   image,
   prompt: 'Identify any safety concerns or defects',
@@ -178,7 +182,7 @@ Extract insights from PDFs with AI:
 const contract = fs.readFileSync('contract.pdf');
 const invoice = fs.readFileSync('invoice.pdf');
 
-// Analyze documents
+// Analyze documents with OpenAI
 const analysis = await ai.openaiProvider.document({
   systemMessage: 'You are a legal expert.',
   userMessage: 'Compare these documents and highlight key differences',
@@ -186,7 +190,7 @@ const analysis = await ai.openaiProvider.document({
   pdfDocuments: [contract, invoice],
 });
 
-// Multi-document analysis
+// Multi-document analysis with Anthropic
 const taxDocs = [form1099, w2, receipts];
 const taxAnalysis = await ai.anthropicProvider.document({
   systemMessage: 'You are a tax advisor.',
@@ -212,6 +216,8 @@ console.log(deepResearch.answer);
 console.log('Sources:', deepResearch.sources);
 
 // Anthropic Web Search - Domain-filtered research
+import { AnthropicProvider } from '@push.rocks/smartai';
+
 const anthropic = new AnthropicProvider({
   anthropicToken: 'sk-ant-...',
   enableWebSearch: true,
@@ -232,14 +238,14 @@ const perplexityResearch = await ai.perplexityProvider.research({
 
 **Research Options:**
 
-- `searchDepth`: 'basic' | 'advanced' | 'deep'
+- `searchDepth`: `'basic'` | `'advanced'` | `'deep'`
 - `maxSources`: Number of sources to include
 - `includeWebSearch`: Enable web search (OpenAI)
 - `background`: Run as background task (OpenAI)
 
 **Supported Providers:**
 
-- **OpenAI**: Deep Research API with specialized models (`o3-deep-research-2025-06-26`, `o4-mini-deep-research-2025-06-26`)
+- **OpenAI**: Deep Research API with specialized models (`o3-deep-research-*`, `o4-mini-deep-research-*`)
 - **Anthropic**: Web Search API with domain filtering
 - **Perplexity**: Sonar and Sonar Pro models with built-in citations
 
@@ -268,12 +274,12 @@ const response = await anthropic.chat({
 
 **Thinking Modes:**
 
-| Mode | Budget Tokens | Use Case |
+| Mode       | Budget Tokens | Use Case                                         |
-|------|---------------|----------|
+| ---------- | ------------- | ------------------------------------------------ |
-| `'quick'` | 2,048 | Lightweight reasoning for simple queries |
+| `'quick'`  | 2,048         | Lightweight reasoning for simple queries         |
-| `'normal'` | 8,000 | **Default** - Balanced reasoning for most tasks |
+| `'normal'` | 8,000         | **Default** — Balanced reasoning for most tasks  |
-| `'deep'` | 16,000 | Complex reasoning for difficult problems |
+| `'deep'`   | 16,000        | Complex reasoning for difficult problems         |
-| `'off'` | 0 | Disable extended thinking |
+| `'off'`    | 0             | Disable extended thinking                        |
 
 **Best Practices:**
 
@@ -282,6 +288,39 @@ const response = await anthropic.chat({
 - Use `'quick'` for simple factual queries where deep reasoning isn't needed
 - Thinking budget counts against total token usage
 
+### 📑 Native PDF OCR (Mistral)
+
+Mistral provides native PDF document processing via their OCR API — no image conversion required:
+
+```typescript
+import { MistralProvider } from '@push.rocks/smartai';
+
+const mistral = new MistralProvider({
+  mistralToken: 'your-api-key',
+  chatModel: 'mistral-large-latest', // Default
+  ocrModel: 'mistral-ocr-latest', // Default
+  tableFormat: 'markdown', // 'markdown' | 'html'
+});
+
+await mistral.start();
+
+// Direct PDF processing - no image conversion overhead
+const result = await mistral.document({
+  systemMessage: 'You are a document analyst.',
+  userMessage: 'Extract all invoice details and calculate the total.',
+  pdfDocuments: [invoicePdfBuffer],
+  messageHistory: [],
+});
+```
+
+**Key Advantage**: Unlike other providers that convert PDFs to images first, Mistral's OCR API processes PDFs natively, potentially offering faster and more accurate text extraction for document-heavy workloads.
+
+**Supported Formats:**
+
+- Native PDF processing via Files API
+- Image OCR (JPEG, PNG, GIF, WebP) for vision tasks
+- Table extraction with markdown or HTML output
+
 ### 🎨 Image Generation & Editing
 
 Generate and edit images with OpenAI's cutting-edge models:
@@ -348,14 +387,14 @@ const editedImage = await ai.openaiProvider.imageEdit({
 
 **Image Generation Options:**
 
-- `model`: 'gpt-image-1' | 'dall-e-3' | 'dall-e-2'
+- `model`: `'gpt-image-1'` | `'dall-e-3'` | `'dall-e-2'`
-- `quality`: 'low' | 'medium' | 'high' | 'auto'
+- `quality`: `'low'` | `'medium'` | `'high'` | `'auto'`
 - `size`: Multiple aspect ratios up to 4096×4096
-- `background`: 'transparent' | 'opaque' | 'auto'
+- `background`: `'transparent'` | `'opaque'` | `'auto'`
-- `outputFormat`: 'png' | 'jpeg' | 'webp'
+- `outputFormat`: `'png'` | `'jpeg'` | `'webp'`
-- `outputCompression`: 0-100 for webp/jpeg
+- `outputCompression`: 0–100 for webp/jpeg
-- `moderation`: 'low' | 'auto'
+- `moderation`: `'low'` | `'auto'`
-- `n`: Number of images (1-10)
+- `n`: Number of images (1–10)
 
 **gpt-image-1 Advantages:**
 
@@ -391,7 +430,7 @@ await inputWriter.write('Now show me how to make it thread-safe');
 
 ```typescript
 const supportBot = new SmartAi({
-  anthropicToken: process.env.ANTHROPIC_KEY // Claude for empathetic responses
+  anthropicToken: process.env.ANTHROPIC_KEY, // Claude for empathetic responses
 });
 
 async function handleCustomerQuery(query: string, history: ChatMessage[]) {
@@ -400,13 +439,13 @@ async function handleCustomerQuery(query: string, history: ChatMessage[]) {
       systemMessage: `You are a helpful customer support agent.
                       Be empathetic, professional, and solution-oriented.`,
       userMessage: query,
-      messageHistory: history
+      messageHistory: history,
     });
 
     return response.message;
   } catch (error) {
     // Fallback to another provider if needed
-    return await supportBot.openaiProvider.chat({...});
+    return await supportBot.openaiProvider.chat({ /* ... */ });
   }
 }
 ```
@@ -419,8 +458,6 @@ const codeReviewer = new SmartAi({
 });
 
 async function reviewCode(code: string, language: string) {
-  const startTime = Date.now();
-
   const review = await codeReviewer.groqProvider.chat({
     systemMessage: `You are a ${language} expert. Review code for:
                     - Security vulnerabilities
@@ -431,7 +468,6 @@ async function reviewCode(code: string, language: string) {
     messageHistory: [],
   });
 
-  console.log(`Review completed in ${Date.now() - startTime}ms`);
   return review.message;
 }
 ```
@@ -445,14 +481,15 @@ const researcher = new SmartAi({
 
 async function research(topic: string) {
   // Perplexity excels at web-aware research
-  const findings = await researcher.perplexityProvider.chat({
+  const findings = await researcher.perplexityProvider.research({
-    systemMessage:
+    query: `Research the latest developments in ${topic}`,
-      'You are a research assistant. Provide factual, cited information.',
+    searchDepth: 'deep',
-    userMessage: `Research the latest developments in ${topic}`,
-    messageHistory: [],
   });
 
-  return findings.message;
+  return {
+    answer: findings.answer,
+    sources: findings.sources,
+  };
 }
 ```
 
@@ -489,23 +526,26 @@ async function analyzeSensitiveDoc(pdfBuffer: Buffer) {
 class SmartAIRouter {
   constructor(private ai: SmartAi) {}
 
-  async query(message: string, requirements: {
+  async query(
-    speed?: boolean;
+    message: string,
-    accuracy?: boolean;
+    requirements: {
-    cost?: boolean;
+      speed?: boolean;
-    privacy?: boolean;
+      accuracy?: boolean;
-  }) {
+      cost?: boolean;
+      privacy?: boolean;
+    }
+  ) {
     if (requirements.privacy) {
-      return this.ai.ollamaProvider.chat({...}); // Local only
+      return this.ai.ollamaProvider.chat({ /* ... */ }); // Local only
     }
     if (requirements.speed) {
-      return this.ai.groqProvider.chat({...}); // 10x faster
+      return this.ai.groqProvider.chat({ /* ... */ }); // 10x faster
     }
     if (requirements.accuracy) {
-      return this.ai.anthropicProvider.chat({...}); // Best reasoning
+      return this.ai.anthropicProvider.chat({ /* ... */ }); // Best reasoning
     }
     // Default fallback
-    return this.ai.openaiProvider.chat({...});
+    return this.ai.openaiProvider.chat({ /* ... */ });
   }
 }
 ```
@@ -516,7 +556,7 @@ class SmartAIRouter {
 // Don't wait for the entire response
 async function streamResponse(userQuery: string) {
   const stream = await ai.openaiProvider.chatStream(
-    createInputStream(userQuery),
+    createInputStream(userQuery)
   );
 
   // Process tokens as they arrive
@@ -533,9 +573,9 @@ async function streamResponse(userQuery: string) {
 // Get the best answer from multiple AIs
 async function consensusQuery(question: string) {
   const providers = [
-    ai.openaiProvider.chat({...}),
+    ai.openaiProvider.chat({ /* ... */ }),
-    ai.anthropicProvider.chat({...}),
+    ai.anthropicProvider.chat({ /* ... */ }),
-    ai.perplexityProvider.chat({...})
+    ai.perplexityProvider.chat({ /* ... */ }),
   ];
 
   const responses = await Promise.all(providers);
@@ -543,21 +583,61 @@ async function consensusQuery(question: string) {
 }
 ```
 
-## 🛠️ Advanced Features
+## 🛠️ Advanced Configuration
 
-### Custom Streaming Transformations
+### Provider-Specific Options
 
 ```typescript
-// Add real-time translation
+const ai = new SmartAi({
-const translationStream = new TransformStream({
+  // OpenAI
-  async transform(chunk, controller) {
+  openaiToken: 'sk-...',
-    const translated = await translateChunk(chunk);
+
-    controller.enqueue(translated);
+  // Anthropic with extended thinking
+  anthropicToken: 'sk-ant-...',
+
+  // Perplexity for research
+  perplexityToken: 'pplx-...',
+
+  // Groq for speed
+  groqToken: 'gsk_...',
+
+  // Mistral with OCR settings
+  mistralToken: 'your-key',
+  mistral: {
+    chatModel: 'mistral-large-latest',
+    ocrModel: 'mistral-ocr-latest',
+    tableFormat: 'markdown',
+  },
+
+  // XAI (Grok)
+  xaiToken: 'xai-...',
+
+  // ElevenLabs TTS
+  elevenlabsToken: 'sk-...',
+  elevenlabs: {
+    defaultVoiceId: '19STyYD15bswVz51nqLf',
+    defaultModelId: 'eleven_v3',
+  },
+
+  // Ollama (local)
+  ollama: {
+    baseUrl: 'http://localhost:11434',
+    model: 'llama2',
+    visionModel: 'llava',
+    defaultOptions: {
+      num_ctx: 4096,
+      temperature: 0.7,
+      top_p: 0.9,
+    },
+    defaultTimeout: 120000,
+  },
+
+  // Exo (distributed)
+  exo: {
+    baseUrl: 'http://localhost:8080/v1',
+    apiKey: 'optional-key',
   },
 });
-
-const responseStream = await ai.openaiProvider.chatStream(input);
-const translatedStream = responseStream.pipeThrough(translationStream);
 ```
 
 ### Error Handling & Fallbacks
@@ -580,83 +660,27 @@ class ResilientAI {
 }
 ```
 
-### Token Counting & Cost Management
-
-```typescript
-// Track usage across providers
-class UsageTracker {
-  async trackedChat(provider: string, options: ChatOptions) {
-    const start = Date.now();
-    const response = await ai[`${provider}Provider`].chat(options);
-
-    const usage = {
-      provider,
-      duration: Date.now() - start,
-      inputTokens: estimateTokens(options),
-      outputTokens: estimateTokens(response.message),
-    };
-
-    await this.logUsage(usage);
-    return response;
-  }
-}
-```
-
-## 📦 Installation & Setup
-
-### Prerequisites
-
-- Node.js 16+
-- TypeScript 4.5+
-- API keys for your chosen providers
-
-### Environment Setup
-
-```bash
-# Install
-npm install @push.rocks/smartai
-
-# Set up environment variables
-export OPENAI_API_KEY=sk-...
-export ANTHROPIC_API_KEY=sk-ant-...
-export PERPLEXITY_API_KEY=pplx-...
-export ELEVENLABS_API_KEY=sk-...
-# ... etc
-```
-
-### TypeScript Configuration
-
-```json
-{
-  "compilerOptions": {
-    "target": "ES2022",
-    "module": "NodeNext",
-    "lib": ["ES2022"],
-    "strict": true,
-    "esModuleInterop": true,
-    "skipLibCheck": true
-  }
-}
-```
-
 ## 🎯 Choosing the Right Provider
 
 | Use Case              | Recommended Provider | Why                                                       |
 | --------------------- | -------------------- | --------------------------------------------------------- |
 | **General Purpose**   | OpenAI               | Most features, stable, well-documented                    |
-| **Complex Reasoning** | Anthropic            | Superior logical thinking, safer outputs                  |
+| **Complex Reasoning** | Anthropic            | Superior logical thinking, extended thinking, safer       |
+| **Document OCR**      | Mistral              | Native PDF processing, no image conversion overhead       |
 | **Research & Facts**  | Perplexity           | Web-aware, provides citations                             |
 | **Deep Research**     | OpenAI               | Deep Research API with comprehensive analysis             |
-| **Premium TTS**       | ElevenLabs           | Most natural voices, 70+ languages, superior quality (v3) |
+| **Premium TTS**       | ElevenLabs           | Most natural voices, 70+ languages, v3 model              |
 | **Speed Critical**    | Groq                 | 10x faster inference, sub-second responses                |
 | **Privacy Critical**  | Ollama               | 100% local, no data leaves your servers                   |
-| **Real-time Data**    | XAI                  | Access to current information                             |
+| **Real-time Data**    | XAI                  | Grok with access to current information                   |
 | **Cost Sensitive**    | Ollama/Exo           | Free (local) or distributed compute                       |
 
 ## 📈 Roadmap
 
 - [x] Research & Web Search API
 - [x] Image generation support (gpt-image-1, DALL-E 3, DALL-E 2)
+- [x] Extended thinking (Anthropic)
+- [x] Native PDF OCR (Mistral)
 - [ ] Streaming function calls
 - [ ] Voice input processing
 - [ ] Fine-tuning integration

test/test.audio.elevenlabs.ts

@@ -1,8 +1,9 @@
 import { expect, tap } from '@git.zone/tstest/tapbundle';
 import * as qenv from '@push.rocks/qenv';
-import * as smartfile from '@push.rocks/smartfile';
+import { SmartFs, SmartFsProviderNode } from '@push.rocks/smartfs';
 
 const testQenv = new qenv.Qenv('./', './.nogit/');
+const smartfs = new SmartFs(new SmartFsProviderNode());
 
 import * as smartai from '../ts/index.js';
 
@@ -27,7 +28,7 @@ tap.test('ElevenLabs Audio: should create audio response', async () => {
     chunks.push(chunk as Uint8Array);
   }
   const audioBuffer = Buffer.concat(chunks);
-  await smartfile.fs.toFs(audioBuffer, './.nogit/testoutput_elevenlabs.mp3');
+  await smartfs.file('./.nogit/testoutput_elevenlabs.mp3').write(audioBuffer);
   console.log(`Audio Buffer length: ${audioBuffer.length}`);
   expect(audioBuffer.length).toBeGreaterThan(0);
 });
@@ -42,7 +43,7 @@ tap.test('ElevenLabs Audio: should create audio with custom voice', async () =>
     chunks.push(chunk as Uint8Array);
   }
   const audioBuffer = Buffer.concat(chunks);
-  await smartfile.fs.toFs(audioBuffer, './.nogit/testoutput_elevenlabs_custom.mp3');
+  await smartfs.file('./.nogit/testoutput_elevenlabs_custom.mp3').write(audioBuffer);
   console.log(`Audio Buffer length (custom voice): ${audioBuffer.length}`);
   expect(audioBuffer.length).toBeGreaterThan(0);
 });

test/test.audio.openai.ts

@@ -1,8 +1,9 @@
 import { expect, tap } from '@git.zone/tstest/tapbundle';
 import * as qenv from '@push.rocks/qenv';
-import * as smartfile from '@push.rocks/smartfile';
+import { SmartFs, SmartFsProviderNode } from '@push.rocks/smartfs';
 
 const testQenv = new qenv.Qenv('./', './.nogit/');
+const smartfs = new SmartFs(new SmartFsProviderNode());
 
 import * as smartai from '../ts/index.js';
 
@@ -26,7 +27,7 @@ tap.test('OpenAI Audio: should create audio response', async () => {
     chunks.push(chunk as Uint8Array);
   }
   const audioBuffer = Buffer.concat(chunks);
-  await smartfile.fs.toFs(audioBuffer, './.nogit/testoutput.mp3');
+  await smartfs.file('./.nogit/testoutput.mp3').write(audioBuffer);
   console.log(`Audio Buffer length: ${audioBuffer.length}`);
   // Assert that the resulting buffer is not empty.
   expect(audioBuffer.length).toBeGreaterThan(0);

test/test.chat.mistral.ts

@@ -0,0 +1,66 @@
+import { expect, tap } from '@git.zone/tstest/tapbundle';
+import * as qenv from '@push.rocks/qenv';
+
+const testQenv = new qenv.Qenv('./', './.nogit/');
+
+import * as smartai from '../ts/index.js';
+
+let mistralProvider: smartai.MistralProvider;
+
+tap.test('Mistral Chat: should create and start Mistral provider', async () => {
+  mistralProvider = new smartai.MistralProvider({
+    mistralToken: await testQenv.getEnvVarOnDemand('MISTRAL_API_KEY'),
+  });
+  await mistralProvider.start();
+  expect(mistralProvider).toBeInstanceOf(smartai.MistralProvider);
+});
+
+tap.test('Mistral Chat: should create chat response', async () => {
+  const userMessage = 'What is the capital of France? Answer in one word.';
+  const response = await mistralProvider.chat({
+    systemMessage: 'You are a helpful assistant. Be concise.',
+    userMessage: userMessage,
+    messageHistory: [],
+  });
+  console.log(`Mistral Chat - User: ${userMessage}`);
+  console.log(`Mistral Chat - Response: ${response.message}`);
+
+  expect(response.role).toEqual('assistant');
+  expect(response.message).toBeTruthy();
+  expect(response.message.toLowerCase()).toInclude('paris');
+});
+
+tap.test('Mistral Chat: should handle message history', async () => {
+  const messageHistory: smartai.ChatMessage[] = [
+    { role: 'user', content: 'My name is Claude Test' },
+    { role: 'assistant', content: 'Nice to meet you, Claude Test!' }
+  ];
+
+  const response = await mistralProvider.chat({
+    systemMessage: 'You are a helpful assistant with good memory.',
+    userMessage: 'What is my name?',
+    messageHistory: messageHistory,
+  });
+
+  console.log(`Mistral Memory Test - Response: ${response.message}`);
+  expect(response.message.toLowerCase()).toInclude('claude test');
+});
+
+tap.test('Mistral Chat: should handle longer conversations', async () => {
+  const response = await mistralProvider.chat({
+    systemMessage: 'You are a helpful coding assistant.',
+    userMessage: 'Write a simple hello world function in TypeScript. Keep it brief.',
+    messageHistory: [],
+  });
+
+  console.log(`Mistral Coding Test - Response: ${response.message}`);
+  expect(response.message).toBeTruthy();
+  // Should contain some TypeScript/function code
+  expect(response.message).toInclude('function');
+});
+
+tap.test('Mistral Chat: should stop the provider', async () => {
+  await mistralProvider.stop();
+});
+
+export default tap.start();

test/test.document.anthropic.ts

@@ -1,9 +1,10 @@
 import { expect, tap } from '@git.zone/tstest/tapbundle';
 import * as qenv from '@push.rocks/qenv';
 import * as smartrequest from '@push.rocks/smartrequest';
-import * as smartfile from '@push.rocks/smartfile';
+import { SmartFs, SmartFsProviderNode } from '@push.rocks/smartfs';
 
 const testQenv = new qenv.Qenv('./', './.nogit/');
+const smartfs = new SmartFs(new SmartFsProviderNode());
 
 import * as smartai from '../ts/index.js';
 
@@ -41,7 +42,7 @@ tap.test('Anthropic Document: should handle complex document analysis', async ()
   let pdfBuffer: Uint8Array;
 
   try {
-    pdfBuffer = await smartfile.fs.toBuffer(pdfPath);
+    pdfBuffer = await smartfs.file(pdfPath).read();
   } catch (error) {
     // If the file doesn't exist, use the dummy PDF
     console.log('Demo PDF not found, using dummy PDF instead');

test/test.document.mistral.ts

@@ -0,0 +1,100 @@
+import { expect, tap } from '@git.zone/tstest/tapbundle';
+import * as qenv from '@push.rocks/qenv';
+import * as smartrequest from '@push.rocks/smartrequest';
+import { SmartFs, SmartFsProviderNode } from '@push.rocks/smartfs';
+
+const testQenv = new qenv.Qenv('./', './.nogit/');
+const smartfs = new SmartFs(new SmartFsProviderNode());
+
+import * as smartai from '../ts/index.js';
+
+let mistralProvider: smartai.MistralProvider;
+
+tap.test('Mistral Document: should create and start Mistral provider', async () => {
+  mistralProvider = new smartai.MistralProvider({
+    mistralToken: await testQenv.getEnvVarOnDemand('MISTRAL_API_KEY'),
+    tableFormat: 'markdown',
+  });
+  await mistralProvider.start();
+  expect(mistralProvider).toBeInstanceOf(smartai.MistralProvider);
+});
+
+tap.test('Mistral Document: should process a PDF document', async () => {
+  const pdfUrl = 'https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf';
+  const pdfResponse = await smartrequest.SmartRequest.create()
+    .url(pdfUrl)
+    .get();
+
+  const result = await mistralProvider.document({
+    systemMessage: 'Classify the document. Only the following answers are allowed: "invoice", "bank account statement", "contract", "test document", "other". The answer should only contain the keyword for machine use.',
+    userMessage: 'Classify this document.',
+    messageHistory: [],
+    pdfDocuments: [Buffer.from(await pdfResponse.arrayBuffer())],
+  });
+
+  console.log(`Mistral Document - Result:`, result);
+  expect(result).toBeTruthy();
+  expect(result.message).toBeTruthy();
+});
+
+tap.test('Mistral Document: should handle complex document analysis', async () => {
+  // Test with the demo PDF if it exists
+  const pdfPath = './.nogit/demo_without_textlayer.pdf';
+  let pdfBuffer: Uint8Array;
+
+  try {
+    pdfBuffer = await smartfs.file(pdfPath).read();
+  } catch (error) {
+    // If the file doesn't exist, use the dummy PDF
+    console.log('Demo PDF not found, using dummy PDF instead');
+    const pdfUrl = 'https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf';
+    const pdfResponse = await smartrequest.SmartRequest.create()
+      .url(pdfUrl)
+      .get();
+    pdfBuffer = Buffer.from(await pdfResponse.arrayBuffer());
+  }
+
+  const result = await mistralProvider.document({
+    systemMessage: `
+      Analyze this document and provide a JSON response with the following structure:
+      {
+        "documentType": "string",
+        "hasText": boolean,
+        "summary": "string"
+      }
+    `,
+    userMessage: 'Analyze this document.',
+    messageHistory: [],
+    pdfDocuments: [pdfBuffer],
+  });
+
+  console.log(`Mistral Complex Document Analysis:`, result);
+  expect(result).toBeTruthy();
+  expect(result.message).toBeTruthy();
+});
+
+tap.test('Mistral Document: should process multiple PDF documents', async () => {
+  const pdfUrl = 'https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf';
+  const pdfResponse = await smartrequest.SmartRequest.create()
+    .url(pdfUrl)
+    .get();
+
+  const pdfBuffer = Buffer.from(await pdfResponse.arrayBuffer());
+
+  const result = await mistralProvider.document({
+    systemMessage: 'You are a document comparison assistant.',
+    userMessage: 'Are these two documents the same? Answer yes or no.',
+    messageHistory: [],
+    pdfDocuments: [pdfBuffer, pdfBuffer], // Same document twice for test
+  });
+
+  console.log(`Mistral Multi-Document - Result:`, result);
+  expect(result).toBeTruthy();
+  expect(result.message).toBeTruthy();
+});
+
+tap.test('Mistral Document: should stop the provider', async () => {
+  await mistralProvider.stop();
+});
+
+export default tap.start();

test/test.document.openai.ts

@@ -1,9 +1,10 @@
 import { expect, tap } from '@git.zone/tstest/tapbundle';
 import * as qenv from '@push.rocks/qenv';
 import * as smartrequest from '@push.rocks/smartrequest';
-import * as smartfile from '@push.rocks/smartfile';
+import { SmartFs, SmartFsProviderNode } from '@push.rocks/smartfs';
 
 const testQenv = new qenv.Qenv('./', './.nogit/');
+const smartfs = new SmartFs(new SmartFsProviderNode());
 
 import * as smartai from '../ts/index.js';
 
@@ -32,7 +33,7 @@ tap.test('OpenAI Document: should document a pdf', async () => {
 });
 
 tap.test('OpenAI Document: should recognize companies in a pdf', async () => {
-  const pdfBuffer = await smartfile.fs.toBuffer('./.nogit/demo_without_textlayer.pdf');
+  const pdfBuffer = await smartfs.file('./.nogit/demo_without_textlayer.pdf').read();
   const result = await testSmartai.openaiProvider.document({
     systemMessage: `
       summarize the document.

test/test.vision.anthropic.ts

@@ -1,8 +1,9 @@
 import { expect, tap } from '@git.zone/tstest/tapbundle';
 import * as qenv from '@push.rocks/qenv';
-import * as smartfile from '@push.rocks/smartfile';
+import { SmartFs, SmartFsProviderNode } from '@push.rocks/smartfs';
 
 const testQenv = new qenv.Qenv('./', './.nogit/');
+const smartfs = new SmartFs(new SmartFsProviderNode());
 
 import * as smartai from '../ts/index.js';
 
@@ -21,7 +22,7 @@ tap.test('Anthropic Vision: should analyze coffee image with latte art', async (
   const imagePath = './test/testimages/coffee-dani/coffee.jpg';
   console.log(`Loading coffee image from: ${imagePath}`);
 
-  const imageBuffer = await smartfile.fs.toBuffer(imagePath);
+  const imageBuffer = await smartfs.file(imagePath).read();
   console.log(`Image loaded, size: ${imageBuffer.length} bytes`);
 
   const result = await anthropicProvider.vision({
@@ -45,7 +46,7 @@ tap.test('Anthropic Vision: should analyze laptop/workspace image', async () =>
   const imagePath = './test/testimages/laptop-nicolas/laptop.jpg';
   console.log(`Loading laptop image from: ${imagePath}`);
 
-  const imageBuffer = await smartfile.fs.toBuffer(imagePath);
+  const imageBuffer = await smartfs.file(imagePath).read();
   console.log(`Image loaded, size: ${imageBuffer.length} bytes`);
 
   const result = await anthropicProvider.vision({
@@ -69,7 +70,7 @@ tap.test('Anthropic Vision: should analyze receipt/document image', async () =>
   const imagePath = './test/testimages/receipt-annie/receipt.jpg';
   console.log(`Loading receipt image from: ${imagePath}`);
 
-  const imageBuffer = await smartfile.fs.toBuffer(imagePath);
+  const imageBuffer = await smartfs.file(imagePath).read();
   console.log(`Image loaded, size: ${imageBuffer.length} bytes`);
 
   const result = await anthropicProvider.vision({

ts/00_commitinfo_data.ts

@@ -3,6 +3,6 @@
  */
 export const commitinfo = {
   name: '@push.rocks/smartai',
-  version: '0.9.0',
+  version: '0.13.3',
   description: 'SmartAi is a versatile TypeScript library designed to facilitate integration and interaction with various AI models, offering functionalities for chat, audio generation, document processing, and vision tasks.'
 }

ts/abstract.classes.multimodal.ts

@@ -6,6 +6,10 @@ import * as plugins from './plugins.js';
 export interface ChatMessage {
   role: 'assistant' | 'user' | 'system';
   content: string;
+  /** Base64-encoded images for vision-capable models */
+  images?: string[];
+  /** Chain-of-thought reasoning for GPT-OSS models (e.g., Ollama) */
+  reasoning?: string;
 }
 
 /**
@@ -15,6 +19,16 @@ export interface ChatOptions {
   systemMessage: string;
   userMessage: string;
   messageHistory: ChatMessage[];
+  /** Base64-encoded images for the current message (vision-capable models) */
+  images?: string[];
+}
+
+/**
+ * Options for streaming chat interactions
+ */
+export interface StreamingChatOptions extends ChatOptions {
+  /** Callback fired for each token during generation */
+  onToken?: (token: string) => void;
 }
 
 /**
@@ -23,6 +37,8 @@ export interface ChatOptions {
 export interface ChatResponse {
   role: 'assistant';
   message: string;
+  /** Chain-of-thought reasoning from reasoning models */
+  reasoning?: string;
 }
 
 /**
@@ -161,6 +177,14 @@ export abstract class MultiModalModel {
    */
   public abstract chatStream(input: ReadableStream<Uint8Array>): Promise<ReadableStream<string>>;
 
+  /**
+   * Streaming chat with token callback
+   * Calls onToken for each token generated, returns final response
+   * @param optionsArg Options containing system message, user message, message history, and onToken callback
+   * @returns Promise resolving to the assistant's response
+   */
+  public chatStreaming?(optionsArg: StreamingChatOptions): Promise<ChatResponse>;
+
   /**
    * Text-to-speech conversion
    * @param optionsArg Options containing the message to convert to speech

ts/classes.conversation.ts

@@ -84,6 +84,18 @@ export class Conversation {
     return conversation;
   }
 
+  public static async createWithMistral(smartaiRefArg: SmartAi) {
+    if (!smartaiRefArg.mistralProvider) {
+      throw new Error('Mistral provider not available');
+    }
+    const conversation = new Conversation(smartaiRefArg, {
+      processFunction: async (input) => {
+        return '' // TODO implement proper streaming
+      }
+    });
+    return conversation;
+  }
+
   public static async createWithXai(smartaiRefArg: SmartAi) {
     if (!smartaiRefArg.xaiProvider) {
       throw new Error('XAI provider not available');

ts/classes.smartai.ts

@@ -2,7 +2,8 @@ import { Conversation } from './classes.conversation.js';
 import * as plugins from './plugins.js';
 import { AnthropicProvider } from './provider.anthropic.js';
 import { ElevenLabsProvider } from './provider.elevenlabs.js';
-import { OllamaProvider } from './provider.ollama.js';
+import { MistralProvider } from './provider.mistral.js';
+import { OllamaProvider, type IOllamaModelOptions } from './provider.ollama.js';
 import { OpenAiProvider } from './provider.openai.js';
 import { PerplexityProvider } from './provider.perplexity.js';
 import { ExoProvider } from './provider.exo.js';
@@ -15,16 +16,24 @@ export interface ISmartAiOptions {
   anthropicToken?: string;
   perplexityToken?: string;
   groqToken?: string;
+  mistralToken?: string;
   xaiToken?: string;
   elevenlabsToken?: string;
   exo?: {
     baseUrl?: string;
     apiKey?: string;
   };
+  mistral?: {
+    chatModel?: string;
+    ocrModel?: string;
+    tableFormat?: 'markdown' | 'html';
+  };
   ollama?: {
     baseUrl?: string;
     model?: string;
     visionModel?: string;
+    defaultOptions?: IOllamaModelOptions;
+    defaultTimeout?: number;
   };
   elevenlabs?: {
     defaultVoiceId?: string;
@@ -32,7 +41,7 @@ export interface ISmartAiOptions {
   };
 }
 
-export type TProvider = 'openai' | 'anthropic' | 'perplexity' | 'ollama' | 'exo' | 'groq' | 'xai' | 'elevenlabs';
+export type TProvider = 'openai' | 'anthropic' | 'perplexity' | 'ollama' | 'exo' | 'groq' | 'mistral' | 'xai' | 'elevenlabs';
 
 export class SmartAi {
   public options: ISmartAiOptions;
@@ -43,6 +52,7 @@ export class SmartAi {
   public ollamaProvider: OllamaProvider;
   public exoProvider: ExoProvider;
   public groqProvider: GroqProvider;
+  public mistralProvider: MistralProvider;
   public xaiProvider: XAIProvider;
   public elevenlabsProvider: ElevenLabsProvider;
 
@@ -75,6 +85,15 @@ export class SmartAi {
       });
       await this.groqProvider.start();
     }
+    if (this.options.mistralToken) {
+      this.mistralProvider = new MistralProvider({
+        mistralToken: this.options.mistralToken,
+        chatModel: this.options.mistral?.chatModel,
+        ocrModel: this.options.mistral?.ocrModel,
+        tableFormat: this.options.mistral?.tableFormat,
+      });
+      await this.mistralProvider.start();
+    }
     if (this.options.xaiToken) {
       this.xaiProvider = new XAIProvider({
         xaiToken: this.options.xaiToken,
@@ -94,6 +113,8 @@ export class SmartAi {
         baseUrl: this.options.ollama.baseUrl,
         model: this.options.ollama.model,
         visionModel: this.options.ollama.visionModel,
+        defaultOptions: this.options.ollama.defaultOptions,
+        defaultTimeout: this.options.ollama.defaultTimeout,
       });
       await this.ollamaProvider.start();
     }
@@ -119,6 +140,9 @@ export class SmartAi {
     if (this.groqProvider) {
       await this.groqProvider.stop();
     }
+    if (this.mistralProvider) {
+      await this.mistralProvider.stop();
+    }
     if (this.xaiProvider) {
       await this.xaiProvider.stop();
     }
@@ -150,6 +174,8 @@ export class SmartAi {
         return Conversation.createWithOllama(this);
       case 'groq':
         return Conversation.createWithGroq(this);
+      case 'mistral':
+        return Conversation.createWithMistral(this);
       case 'xai':
         return Conversation.createWithXai(this);
       case 'elevenlabs':

ts/index.ts

@@ -4,6 +4,7 @@ export * from './provider.openai.js';
 export * from './provider.anthropic.js';
 export * from './provider.perplexity.js';
 export * from './provider.groq.js';
+export * from './provider.mistral.js';
 export * from './provider.ollama.js';
 export * from './provider.xai.js';
 export * from './provider.exo.js';

ts/plugins.ts

@@ -8,7 +8,7 @@ export {
 // @push.rocks scope
 import * as qenv from '@push.rocks/qenv';
 import * as smartarray from '@push.rocks/smartarray';
-import * as smartfile from '@push.rocks/smartfile';
+import * as smartfs from '@push.rocks/smartfs';
 import * as smartpath from '@push.rocks/smartpath';
 import * as smartpdf from '@push.rocks/smartpdf';
 import * as smartpromise from '@push.rocks/smartpromise';
@@ -18,7 +18,7 @@ import * as webstream from '@push.rocks/webstream';
 export {
   smartarray,
   qenv,
-  smartfile,
+  smartfs,
   smartpath,
   smartpdf,
   smartpromise,
@@ -28,9 +28,11 @@ export {
 
 // third party
 import * as anthropic from '@anthropic-ai/sdk';
+import * as mistralai from '@mistralai/mistralai';
 import * as openai from 'openai';
 
 export {
   anthropic,
+  mistralai,
   openai,
 }

ts/provider.mistral.ts

@@ -0,0 +1,352 @@
+import * as plugins from './plugins.js';
+import { MultiModalModel } from './abstract.classes.multimodal.js';
+import type {
+  ChatOptions,
+  ChatResponse,
+  ChatMessage,
+  ResearchOptions,
+  ResearchResponse,
+  ImageGenerateOptions,
+  ImageEditOptions,
+  ImageResponse
+} from './abstract.classes.multimodal.js';
+
+export interface IMistralProviderOptions {
+  mistralToken: string;
+  chatModel?: string;       // default: 'mistral-large-latest'
+  ocrModel?: string;        // default: 'mistral-ocr-latest'
+  tableFormat?: 'markdown' | 'html';
+}
+
+export class MistralProvider extends MultiModalModel {
+  private options: IMistralProviderOptions;
+  public mistralClient: plugins.mistralai.Mistral;
+
+  constructor(optionsArg: IMistralProviderOptions) {
+    super();
+    this.options = optionsArg;
+  }
+
+  async start() {
+    await super.start();
+    this.mistralClient = new plugins.mistralai.Mistral({
+      apiKey: this.options.mistralToken,
+    });
+  }
+
+  async stop() {
+    await super.stop();
+  }
+
+  /**
+   * Synchronous chat interaction using Mistral's chat API
+   */
+  public async chat(optionsArg: ChatOptions): Promise<ChatResponse> {
+    // Convert message history to Mistral format
+    const messages: Array<{
+      role: 'system' | 'user' | 'assistant';
+      content: string;
+    }> = [];
+
+    // Add system message first
+    if (optionsArg.systemMessage) {
+      messages.push({
+        role: 'system',
+        content: optionsArg.systemMessage
+      });
+    }
+
+    // Add message history
+    for (const msg of optionsArg.messageHistory) {
+      messages.push({
+        role: msg.role === 'system' ? 'system' : msg.role === 'assistant' ? 'assistant' : 'user',
+        content: msg.content
+      });
+    }
+
+    // Add current user message
+    messages.push({
+      role: 'user',
+      content: optionsArg.userMessage
+    });
+
+    const result = await this.mistralClient.chat.complete({
+      model: this.options.chatModel || 'mistral-large-latest',
+      messages: messages,
+    });
+
+    // Extract content from response
+    const choice = result.choices?.[0];
+    let content = '';
+
+    if (choice?.message?.content) {
+      if (typeof choice.message.content === 'string') {
+        content = choice.message.content;
+      } else if (Array.isArray(choice.message.content)) {
+        // Handle array of content chunks
+        content = choice.message.content
+          .map((chunk: any) => {
+            if (typeof chunk === 'string') return chunk;
+            if (chunk && typeof chunk === 'object' && 'text' in chunk) return chunk.text;
+            return '';
+          })
+          .join('');
+      }
+    }
+
+    return {
+      role: 'assistant',
+      message: content,
+    };
+  }
+
+  /**
+   * Streaming chat using Mistral's streaming API
+   */
+  public async chatStream(input: ReadableStream<Uint8Array>): Promise<ReadableStream<string>> {
+    const decoder = new TextDecoder();
+    let buffer = '';
+    const mistralClient = this.mistralClient;
+    const chatModel = this.options.chatModel || 'mistral-large-latest';
+
+    const transform = new TransformStream<Uint8Array, string>({
+      async transform(chunk, controller) {
+        buffer += decoder.decode(chunk, { stream: true });
+
+        // Try to parse complete JSON messages from the buffer
+        while (true) {
+          const newlineIndex = buffer.indexOf('\n');
+          if (newlineIndex === -1) break;
+
+          const line = buffer.slice(0, newlineIndex);
+          buffer = buffer.slice(newlineIndex + 1);
+
+          if (line.trim()) {
+            try {
+              const message = JSON.parse(line);
+
+              // Build messages array
+              const messages: Array<{
+                role: 'system' | 'user' | 'assistant';
+                content: string;
+              }> = [];
+
+              if (message.systemMessage) {
+                messages.push({
+                  role: 'system',
+                  content: message.systemMessage
+                });
+              }
+
+              messages.push({
+                role: message.role === 'assistant' ? 'assistant' : 'user',
+                content: message.content
+              });
+
+              // Use Mistral streaming
+              const stream = await mistralClient.chat.stream({
+                model: chatModel,
+                messages: messages,
+              });
+
+              // Process streaming events
+              for await (const event of stream) {
+                const delta = event.data?.choices?.[0]?.delta;
+                if (delta?.content) {
+                  if (typeof delta.content === 'string') {
+                    controller.enqueue(delta.content);
+                  } else if (Array.isArray(delta.content)) {
+                    for (const chunk of delta.content) {
+                      if (typeof chunk === 'string') {
+                        controller.enqueue(chunk);
+                      } else if (chunk && typeof chunk === 'object' && 'text' in chunk) {
+                        controller.enqueue((chunk as any).text);
+                      }
+                    }
+                  }
+                }
+              }
+            } catch (e) {
+              console.error('Failed to parse message:', e);
+            }
+          }
+        }
+      },
+
+      flush(controller) {
+        if (buffer.trim()) {
+          try {
+            const message = JSON.parse(buffer);
+            controller.enqueue(message.content || '');
+          } catch (e) {
+            console.error('Failed to parse remaining buffer:', e);
+          }
+        }
+      }
+    });
+
+    return input.pipeThrough(transform);
+  }
+
+  /**
+   * Audio generation is not supported by Mistral
+   */
+  public async audio(optionsArg: { message: string }): Promise<NodeJS.ReadableStream> {
+    throw new Error('Audio generation is not supported by Mistral. Please use ElevenLabs or OpenAI provider for audio generation.');
+  }
+
+  /**
+   * Vision using Mistral's OCR API for image analysis
+   */
+  public async vision(optionsArg: { image: Buffer; prompt: string }): Promise<string> {
+    const base64Image = optionsArg.image.toString('base64');
+
+    // Detect image type from buffer header
+    let mimeType = 'image/jpeg';
+    if (optionsArg.image[0] === 0x89 && optionsArg.image[1] === 0x50) {
+      mimeType = 'image/png';
+    } else if (optionsArg.image[0] === 0x47 && optionsArg.image[1] === 0x49) {
+      mimeType = 'image/gif';
+    } else if (optionsArg.image[0] === 0x52 && optionsArg.image[1] === 0x49) {
+      mimeType = 'image/webp';
+    }
+
+    // Use OCR API with image data URL
+    const ocrResult = await this.mistralClient.ocr.process({
+      model: this.options.ocrModel || 'mistral-ocr-latest',
+      document: {
+        imageUrl: `data:${mimeType};base64,${base64Image}`,
+        type: 'image_url',
+      },
+    });
+
+    // Combine markdown from all pages
+    const extractedText = ocrResult.pages.map(page => page.markdown).join('\n\n');
+
+    // If a prompt is provided, use chat to analyze the extracted text
+    if (optionsArg.prompt && optionsArg.prompt.trim()) {
+      const chatResponse = await this.chat({
+        systemMessage: 'You are an assistant analyzing image content. The following is text extracted from an image using OCR.',
+        userMessage: `${optionsArg.prompt}\n\nExtracted content:\n${extractedText}`,
+        messageHistory: [],
+      });
+      return chatResponse.message;
+    }
+
+    return extractedText;
+  }
+
+  /**
+   * Document processing using Mistral's OCR API
+   * PDFs are uploaded via Files API first, then processed with OCR
+   */
+  public async document(optionsArg: {
+    systemMessage: string;
+    userMessage: string;
+    pdfDocuments: Uint8Array[];
+    messageHistory: ChatMessage[];
+  }): Promise<{ message: any }> {
+    const extractedTexts: string[] = [];
+    const uploadedFileIds: string[] = [];
+
+    try {
+      // Process each PDF document using Mistral OCR
+      for (let i = 0; i < optionsArg.pdfDocuments.length; i++) {
+        const pdfDocument = optionsArg.pdfDocuments[i];
+
+        // Upload the PDF to Mistral's Files API first
+        const uploadResult = await this.mistralClient.files.upload({
+          file: {
+            fileName: `document_${i + 1}.pdf`,
+            content: pdfDocument,
+          },
+          purpose: 'ocr',
+        });
+
+        uploadedFileIds.push(uploadResult.id);
+
+        // Now use OCR with the uploaded file
+        const ocrResult = await this.mistralClient.ocr.process({
+          model: this.options.ocrModel || 'mistral-ocr-latest',
+          document: {
+            type: 'file',
+            fileId: uploadResult.id,
+          },
+          tableFormat: this.options.tableFormat || 'markdown',
+        });
+
+        // Combine all page markdown with page separators
+        const pageTexts = ocrResult.pages.map((page, index) => {
+          let pageContent = `--- Page ${index + 1} ---\n${page.markdown}`;
+
+          // Include tables if present
+          if (page.tables && page.tables.length > 0) {
+            pageContent += '\n\n**Tables:**\n' + page.tables.map((t: any) => t.markdown || t.html || '').join('\n');
+          }
+
+          // Include header/footer if present
+          if (page.header) {
+            pageContent = `Header: ${page.header}\n${pageContent}`;
+          }
+          if (page.footer) {
+            pageContent += `\nFooter: ${page.footer}`;
+          }
+
+          return pageContent;
+        }).join('\n\n');
+
+        extractedTexts.push(pageTexts);
+      }
+
+      // Combine all document texts
+      const allDocumentText = extractedTexts.length === 1
+        ? extractedTexts[0]
+        : extractedTexts.map((text, i) => `=== Document ${i + 1} ===\n${text}`).join('\n\n');
+
+      // Use chat API to process the extracted text with the user's query
+      const chatResponse = await this.chat({
+        systemMessage: optionsArg.systemMessage || 'You are a helpful assistant analyzing document content.',
+        userMessage: `${optionsArg.userMessage}\n\n---\nDocument Content:\n${allDocumentText}`,
+        messageHistory: optionsArg.messageHistory,
+      });
+
+      return {
+        message: {
+          role: 'assistant',
+          content: chatResponse.message
+        }
+      };
+    } finally {
+      // Clean up uploaded files
+      for (const fileId of uploadedFileIds) {
+        try {
+          await this.mistralClient.files.delete({ fileId });
+        } catch (cleanupError) {
+          // Ignore cleanup errors - files may have already been auto-deleted
+          console.warn(`Failed to delete temporary file ${fileId}:`, cleanupError);
+        }
+      }
+    }
+  }
+
+  /**
+   * Research is not natively supported by Mistral
+   */
+  public async research(optionsArg: ResearchOptions): Promise<ResearchResponse> {
+    throw new Error('Research/web search is not supported by Mistral. Please use Perplexity or Anthropic provider for research capabilities.');
+  }
+
+  /**
+   * Image generation is not supported by Mistral
+   */
+  public async imageGenerate(optionsArg: ImageGenerateOptions): Promise<ImageResponse> {
+    throw new Error('Image generation is not supported by Mistral. Please use OpenAI provider for image generation.');
+  }
+
+  /**
+   * Image editing is not supported by Mistral
+   */
+  public async imageEdit(optionsArg: ImageEditOptions): Promise<ImageResponse> {
+    throw new Error('Image editing is not supported by Mistral. Please use OpenAI provider for image editing.');
+  }
+}

ts/provider.ollama.ts

@@ -9,13 +9,101 @@ import type {
   ResearchResponse,
   ImageGenerateOptions,
   ImageEditOptions,
-  ImageResponse
+  ImageResponse,
+  StreamingChatOptions
 } from './abstract.classes.multimodal.js';
 
+/**
+ * Ollama model runtime options
+ * @see https://github.com/ollama/ollama/blob/main/docs/modelfile.md
+ */
+export interface IOllamaModelOptions {
+  num_ctx?: number;       // Context window (default: 2048)
+  temperature?: number;   // 0 = deterministic (default: 0.8)
+  top_k?: number;         // Top-k sampling (default: 40)
+  top_p?: number;         // Nucleus sampling (default: 0.9)
+  repeat_penalty?: number;// Repeat penalty (default: 1.1)
+  num_predict?: number;   // Max tokens to predict
+  stop?: string[];        // Stop sequences
+  seed?: number;          // Random seed for reproducibility
+  think?: boolean;        // Enable thinking/reasoning mode (for GPT-OSS, QwQ, etc.)
+}
+
+/**
+ * JSON Schema tool definition for Ollama native tool calling
+ * @see https://docs.ollama.com/capabilities/tool-calling
+ */
+export interface IOllamaTool {
+  type: 'function';
+  function: {
+    name: string;
+    description: string;
+    parameters: {
+      type: 'object';
+      properties: Record<string, {
+        type: string;
+        description?: string;
+        enum?: string[];
+      }>;
+      required?: string[];
+    };
+  };
+}
+
+/**
+ * Tool call returned by model in native tool calling mode
+ */
+export interface IOllamaToolCall {
+  function: {
+    name: string;
+    arguments: Record<string, unknown>;
+    index?: number;
+  };
+}
+
 export interface IOllamaProviderOptions {
   baseUrl?: string;
   model?: string;
   visionModel?: string; // Model to use for vision tasks (e.g. 'llava')
+  defaultOptions?: IOllamaModelOptions;  // Default model options
+  defaultTimeout?: number;               // Default timeout in ms (default: 120000)
+}
+
+/**
+ * Extended chat options with Ollama-specific settings
+ */
+export interface IOllamaChatOptions extends ChatOptions {
+  options?: IOllamaModelOptions;  // Per-request model options
+  timeout?: number;               // Per-request timeout in ms
+  model?: string;                 // Per-request model override
+  tools?: IOllamaTool[];          // Available tools for native function calling
+  // images is inherited from ChatOptions
+}
+
+/**
+ * Chunk emitted during streaming
+ */
+export interface IOllamaStreamChunk {
+  content: string;
+  thinking?: string;  // For models with extended thinking
+  toolCalls?: IOllamaToolCall[];  // Tool calls in streaming mode
+  done: boolean;
+  stats?: {
+    totalDuration?: number;
+    evalCount?: number;
+  };
+}
+
+/**
+ * Extended chat response with Ollama-specific fields
+ */
+export interface IOllamaChatResponse extends ChatResponse {
+  thinking?: string;
+  toolCalls?: IOllamaToolCall[];  // Tool calls from model (native tool calling)
+  stats?: {
+    totalDuration?: number;
+    evalCount?: number;
+  };
 }
 
 export class OllamaProvider extends MultiModalModel {
@@ -23,6 +111,8 @@ export class OllamaProvider extends MultiModalModel {
   private baseUrl: string;
   private model: string;
   private visionModel: string;
+  private defaultOptions: IOllamaModelOptions;
+  private defaultTimeout: number;
 
   constructor(optionsArg: IOllamaProviderOptions = {}) {
     super();
@@ -30,6 +120,8 @@ export class OllamaProvider extends MultiModalModel {
     this.baseUrl = optionsArg.baseUrl || 'http://localhost:11434';
     this.model = optionsArg.model || 'llama2';
     this.visionModel = optionsArg.visionModel || 'llava';
+    this.defaultOptions = optionsArg.defaultOptions || {};
+    this.defaultTimeout = optionsArg.defaultTimeout || 120000;
   }
 
   async start() {
@@ -148,23 +240,56 @@ export class OllamaProvider extends MultiModalModel {
   // Implementing the synchronous chat interaction
   public async chat(optionsArg: ChatOptions): Promise<ChatResponse> {
     // Format messages for Ollama
+    const historyMessages = optionsArg.messageHistory.map((msg) => {
+      const formatted: { role: string; content: string; images?: string[]; reasoning?: string } = {
+        role: msg.role,
+        content: msg.content,
+      };
+      if (msg.images && msg.images.length > 0) {
+        formatted.images = msg.images;
+      }
+      if (msg.reasoning) {
+        formatted.reasoning = msg.reasoning;
+      }
+      return formatted;
+    });
+
+    // Build user message with optional images
+    const userMessage: { role: string; content: string; images?: string[] } = {
+      role: 'user',
+      content: optionsArg.userMessage,
+    };
+    if (optionsArg.images && optionsArg.images.length > 0) {
+      userMessage.images = optionsArg.images;
+    }
+
     const messages = [
       { role: 'system', content: optionsArg.systemMessage },
-      ...optionsArg.messageHistory,
+      ...historyMessages,
-      { role: 'user', content: optionsArg.userMessage }
+      userMessage,
     ];
 
-    // Make API call to Ollama
+    // Build request body - include think parameter if set
+    const requestBody: Record<string, unknown> = {
+      model: this.model,
+      messages: messages,
+      stream: false,
+      options: this.defaultOptions,
+    };
+
+    // Add think parameter for reasoning models (GPT-OSS, QwQ, etc.)
+    if (this.defaultOptions.think !== undefined) {
+      requestBody.think = this.defaultOptions.think;
+    }
+
+    // Make API call to Ollama with defaultOptions and timeout
     const response = await fetch(`${this.baseUrl}/api/chat`, {
       method: 'POST',
       headers: {
         'Content-Type': 'application/json',
       },
-      body: JSON.stringify({
+      body: JSON.stringify(requestBody),
-        model: this.model,
+      signal: AbortSignal.timeout(this.defaultTimeout),
-        messages: messages,
-        stream: false
-      }),
     });
 
     if (!response.ok) {
@@ -176,6 +301,302 @@ export class OllamaProvider extends MultiModalModel {
     return {
       role: 'assistant' as const,
       message: result.message.content,
+      reasoning: result.message.thinking || result.message.reasoning,
+    };
+  }
+
+  /**
+   * Streaming chat with token callback (implements MultiModalModel interface)
+   * Calls onToken for each token generated during the response
+   */
+  public async chatStreaming(optionsArg: StreamingChatOptions): Promise<ChatResponse> {
+    const onToken = optionsArg.onToken;
+
+    // Use existing collectStreamResponse with callback, including images
+    const response = await this.collectStreamResponse(
+      {
+        systemMessage: optionsArg.systemMessage,
+        userMessage: optionsArg.userMessage,
+        messageHistory: optionsArg.messageHistory,
+        images: optionsArg.images,
+      },
+      (chunk) => {
+        if (onToken) {
+          if (chunk.thinking) onToken(chunk.thinking);
+          if (chunk.content) onToken(chunk.content);
+        }
+      }
+    );
+
+    return {
+      role: 'assistant' as const,
+      message: response.message,
+      reasoning: response.thinking,
+    };
+  }
+
+  /**
+   * Streaming chat with async iteration and options support
+   */
+  public async chatStreamResponse(
+    optionsArg: IOllamaChatOptions
+  ): Promise<AsyncIterable<IOllamaStreamChunk>> {
+    const model = optionsArg.model || this.model;
+    const timeout = optionsArg.timeout || this.defaultTimeout;
+    const modelOptions = { ...this.defaultOptions, ...optionsArg.options };
+
+    // Format history messages with optional images, reasoning, and tool_calls
+    const historyMessages = optionsArg.messageHistory.map((msg) => {
+      const formatted: { role: string; content: string; images?: string[]; reasoning?: string; tool_calls?: any[] } = {
+        role: msg.role,
+        content: msg.content,
+      };
+      if (msg.images && msg.images.length > 0) {
+        formatted.images = msg.images;
+      }
+      if (msg.reasoning) {
+        formatted.reasoning = msg.reasoning;
+      }
+      // CRITICAL: Include tool_calls in history for native tool calling
+      // Without this, the model doesn't know it already called a tool and may call it again
+      if ((msg as any).tool_calls && Array.isArray((msg as any).tool_calls)) {
+        formatted.tool_calls = (msg as any).tool_calls;
+      }
+      return formatted;
+    });
+
+    // Build user message with optional images
+    const userMessage: { role: string; content: string; images?: string[] } = {
+      role: 'user',
+      content: optionsArg.userMessage,
+    };
+    if (optionsArg.images && optionsArg.images.length > 0) {
+      userMessage.images = optionsArg.images;
+    }
+
+    const messages = [
+      { role: 'system', content: optionsArg.systemMessage },
+      ...historyMessages,
+      userMessage,
+    ];
+
+    // Build request body with optional tools and think parameters
+    const requestBody: Record<string, unknown> = {
+      model,
+      messages,
+      stream: true,
+      options: modelOptions,
+    };
+
+    // Add think parameter for reasoning models (GPT-OSS, QwQ, etc.)
+    if (modelOptions.think !== undefined) {
+      requestBody.think = modelOptions.think;
+    }
+
+    // Add tools for native function calling
+    if (optionsArg.tools && optionsArg.tools.length > 0) {
+      requestBody.tools = optionsArg.tools;
+    }
+
+    const response = await fetch(`${this.baseUrl}/api/chat`, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(requestBody),
+      signal: AbortSignal.timeout(timeout),
+    });
+
+    if (!response.ok) {
+      throw new Error(`Ollama API error: ${response.status}`);
+    }
+
+    const reader = response.body!.getReader();
+    const decoder = new TextDecoder();
+
+    return {
+      [Symbol.asyncIterator]: async function* () {
+        let buffer = '';
+        try {
+          while (true) {
+            const { done, value } = await reader.read();
+            if (done) break;
+            buffer += decoder.decode(value, { stream: true });
+            const lines = buffer.split('\n');
+            buffer = lines.pop() || '';
+            for (const line of lines) {
+              if (!line.trim()) continue;
+              try {
+                const json = JSON.parse(line);
+
+                // Parse tool_calls from response
+                let toolCalls: IOllamaToolCall[] | undefined;
+                if (json.message?.tool_calls && Array.isArray(json.message.tool_calls)) {
+                  toolCalls = json.message.tool_calls.map((tc: any) => ({
+                    function: {
+                      name: tc.function?.name || '',
+                      arguments: typeof tc.function?.arguments === 'string'
+                        ? JSON.parse(tc.function.arguments)
+                        : tc.function?.arguments || {},
+                      index: tc.index,
+                    },
+                  }));
+                }
+
+                yield {
+                  content: json.message?.content || '',
+                  thinking: json.message?.thinking,
+                  toolCalls,
+                  done: json.done || false,
+                  stats: json.done ? {
+                    totalDuration: json.total_duration,
+                    evalCount: json.eval_count,
+                  } : undefined,
+                } as IOllamaStreamChunk;
+              } catch { /* skip malformed */ }
+            }
+          }
+        } finally {
+          reader.releaseLock();
+        }
+      }
+    };
+  }
+
+  /**
+   * Stream and collect full response with optional progress callback
+   */
+  public async collectStreamResponse(
+    optionsArg: IOllamaChatOptions,
+    onChunk?: (chunk: IOllamaStreamChunk) => void
+  ): Promise<IOllamaChatResponse> {
+    const stream = await this.chatStreamResponse(optionsArg);
+    let content = '';
+    let thinking = '';
+    let toolCalls: IOllamaToolCall[] = [];
+    let stats: IOllamaChatResponse['stats'];
+
+    for await (const chunk of stream) {
+      if (chunk.content) content += chunk.content;
+      if (chunk.thinking) thinking += chunk.thinking;
+      if (chunk.toolCalls) toolCalls = toolCalls.concat(chunk.toolCalls);
+      if (chunk.stats) stats = chunk.stats;
+      if (onChunk) onChunk(chunk);
+    }
+
+    return {
+      role: 'assistant' as const,
+      message: content,
+      thinking: thinking || undefined,
+      toolCalls: toolCalls.length > 0 ? toolCalls : undefined,
+      stats,
+    };
+  }
+
+  /**
+   * Non-streaming chat with full options support
+   */
+  public async chatWithOptions(optionsArg: IOllamaChatOptions): Promise<IOllamaChatResponse> {
+    const model = optionsArg.model || this.model;
+    const timeout = optionsArg.timeout || this.defaultTimeout;
+    const modelOptions = { ...this.defaultOptions, ...optionsArg.options };
+
+    // Format history messages with optional images, reasoning, tool_calls, and tool role
+    const historyMessages = optionsArg.messageHistory.map((msg) => {
+      // Handle tool result messages
+      if ((msg as any).role === 'tool') {
+        return {
+          role: 'tool',
+          content: msg.content,
+          tool_name: (msg as any).toolName,
+        };
+      }
+
+      const formatted: { role: string; content: string; images?: string[]; reasoning?: string; tool_calls?: any[] } = {
+        role: msg.role,
+        content: msg.content,
+      };
+      if (msg.images && msg.images.length > 0) {
+        formatted.images = msg.images;
+      }
+      if (msg.reasoning) {
+        formatted.reasoning = msg.reasoning;
+      }
+      // CRITICAL: Include tool_calls in history for native tool calling
+      // Without this, the model doesn't know it already called a tool and may call it again
+      if ((msg as any).tool_calls && Array.isArray((msg as any).tool_calls)) {
+        formatted.tool_calls = (msg as any).tool_calls;
+      }
+      return formatted;
+    });
+
+    // Build user message with optional images
+    const userMessage: { role: string; content: string; images?: string[] } = {
+      role: 'user',
+      content: optionsArg.userMessage,
+    };
+    if (optionsArg.images && optionsArg.images.length > 0) {
+      userMessage.images = optionsArg.images;
+    }
+
+    const messages = [
+      { role: 'system', content: optionsArg.systemMessage },
+      ...historyMessages,
+      userMessage,
+    ];
+
+    // Build request body with optional tools and think parameters
+    const requestBody: Record<string, unknown> = {
+      model,
+      messages,
+      stream: false,
+      options: modelOptions,
+    };
+
+    // Add think parameter for reasoning models (GPT-OSS, QwQ, etc.)
+    if (modelOptions.think !== undefined) {
+      requestBody.think = modelOptions.think;
+    }
+
+    // Add tools for native function calling
+    if (optionsArg.tools && optionsArg.tools.length > 0) {
+      requestBody.tools = optionsArg.tools;
+    }
+
+    const response = await fetch(`${this.baseUrl}/api/chat`, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(requestBody),
+      signal: AbortSignal.timeout(timeout),
+    });
+
+    if (!response.ok) {
+      throw new Error(`Ollama API error: ${response.statusText}`);
+    }
+
+    const result = await response.json();
+
+    // Parse tool_calls from response
+    let toolCalls: IOllamaToolCall[] | undefined;
+    if (result.message?.tool_calls && Array.isArray(result.message.tool_calls)) {
+      toolCalls = result.message.tool_calls.map((tc: any) => ({
+        function: {
+          name: tc.function?.name || '',
+          arguments: typeof tc.function?.arguments === 'string'
+            ? JSON.parse(tc.function.arguments)
+            : tc.function?.arguments || {},
+          index: tc.index,
+        },
+      }));
+    }
+
+    return {
+      role: 'assistant' as const,
+      message: result.message.content || '',
+      thinking: result.message.thinking,
+      toolCalls,
+      stats: {
+        totalDuration: result.total_duration,
+        evalCount: result.eval_count,
+      },
     };
   }