Update model versions to latest (Claude Sonnet 4.5, GPT-5.1, Grok 4.1, etc.) (#431)

SannidhyaSah · Sannidhya · web-flow · commit c9224d711e99 · 2025-12-02T10:18:21.000-07:00
Co-authored-by: Sannidhya &lt;sann@Sannidhyas-MacBook-Pro.local&gt;
diff --git a/docs/getting-started/connecting-api-provider.md b/docs/getting-started/connecting-api-provider.md
@@ -15,7 +15,7 @@ import KangarooIcon from '@site/src/components/KangarooIcon';
 
 # Connecting Your First AI Provider
 
-Roo Code requires an API key from an AI model provider to function. We recommend these options for accessing the powerful **Claude 3.7 Sonnet** model:
+Roo Code requires an API key from an AI model provider to function. We recommend these options for accessing the powerful **Claude Sonnet 4.5** model:
 
 - **OpenRouter (Recommended):** Provides access to multiple AI models through a single API key. Ideal for getting started quickly with minimal setup. [View pricing](https://openrouter.ai/models?order=pricing-low-to-high).
 - **Anthropic:** Direct access to Claude models. Requires API access approval and may have [rate limits depending on your tier](https://docs.anthropic.com/en/api/rate-limits#requirements-to-advance-tier). See [Anthropic's pricing page](https://www.anthropic.com/pricing#anthropic-api) for details.
@@ -86,11 +86,11 @@ Once you have your API key:
 2. In the welcome screen, select your API provider from the dropdown
 3. Paste your API key into the appropriate field
 4. Select your model:
-   - For **OpenRouter**: select `anthropic/claude-3.7-sonnet` ([model details](https://openrouter.ai/anthropic/claude-3.7-sonnet))
-   - For **Anthropic**: select `claude-3-7-sonnet-20250219` ([model details](https://www.anthropic.com/pricing#anthropic-api))
+   - For **OpenRouter**: select `anthropic/claude-sonnet-4-5` ([model details](https://openrouter.ai/anthropic/claude-sonnet-4-5))
+   - For **Anthropic**: select `claude-sonnet-4-5` ([model details](https://www.anthropic.com/pricing#anthropic-api))
 
 :::info Model Selection Advice
-We strongly recommend **Claude 3.7 Sonnet** for the best experience—it generally "just works" out of the box. Roo Code has been extensively optimized for this model's capabilities and instruction-following behavior.
+We strongly recommend **Claude Sonnet 4.5** for the best experience—it generally "just works" out of the box. Roo Code has been extensively optimized for this model's capabilities and instruction-following behavior.
 
 Selecting alternative models is an advanced feature that introduces complexity. Different models vary significantly in how they follow tool instructions, parse formats, and maintain context through multi-step operations. If you do experiment with other models, choose ones specifically designed for structured reasoning and tool use.
 :::
diff --git a/docs/providers/anthropic.md b/docs/providers/anthropic.md
@@ -36,9 +36,11 @@ Anthropic is an AI safety and research company that builds reliable, interpretab
 
 Roo Code supports the following Anthropic Claude models:
 
+*   `claude-sonnet-4-5` (Latest, Recommended)
+*   `claude-opus-4-5-20251101`
 *   `claude-opus-4-1-20250805`
 *   `claude-opus-4-20250514`
-*   `claude-sonnet-4-20250514` (Recommended)
+*   `claude-sonnet-4-20250514`
 *   `claude-3-7-sonnet-20250219`
 *   `claude-3-7-sonnet-20250219:thinking` (Extended Thinking variant)
 *   `claude-3-5-sonnet-20241022`
diff --git a/docs/providers/bedrock.md b/docs/providers/bedrock.md
@@ -61,6 +61,7 @@ Roo Code supports the following models through Bedrock (based on source code):
     *   `amazon.titan-text-embeddings-v1:0`
     *   `amazon.titan-text-embeddings-v2:0`
 *   **Anthropic:**
+    *   `anthropic.claude-sonnet-4-5-20250929-v1:0` (Default)
     *   `anthropic.claude-opus-4.1-20250514-v1:0`
     *   `anthropic.claude-opus-4-20250514-v1:0`
     *   `anthropic.claude-sonnet-4-20250514-v1:0`
@@ -88,6 +89,9 @@ Roo Code supports the following models through Bedrock (based on source code):
     *   `meta.llama3-1-8b-instruct-v1:0`
     *   `meta.llama3-70b-instruct-v1:0`
     *   `meta.llama3-8b-instruct-v1:0`
+*   **OpenAI:**
+    *   `openai.gpt-oss-20b-1:0`
+    *   `openai.gpt-oss-120b-1:0`
 
 Refer to the [Amazon Bedrock documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html) for the most up-to-date list of available models and their IDs. Make sure to use the *model ID* when configuring Roo Code, not the model name.
 
diff --git a/docs/providers/deepseek.md b/docs/providers/deepseek.md
@@ -36,7 +36,6 @@ Roo Code supports the following DeepSeek models:
 
 *   `deepseek-chat` (Recommended for coding tasks)
 *	`deepseek-reasoner` (Recommended for reasoning tasks)
-*   `deepseek-r1`
 
 ---
 
diff --git a/docs/providers/fireworks.md b/docs/providers/fireworks.md
@@ -43,9 +43,9 @@ Fireworks AI is a high-performance platform for running AI models, offering fast
 Fireworks AI offers several high-performance models:
 
 ### Kimi K2
-* `accounts/fireworks/models/kimi-k2-instruct` (Default)
+* `accounts/fireworks/models/kimi-k2-instruct-0905` (Default)
   - 1 trillion total parameters with 32B activated params
-  - 128K context window
+  - 256K context window
   - Optimized for agentic capabilities
   - $0.60/M input, $2.50/M output
 
@@ -72,6 +72,18 @@ Fireworks AI offers several high-performance models:
   - 671B total parameters with 37B activated
   - $0.90/M input, $0.90/M output
 
+### MiniMax
+* `accounts/fireworks/models/minimax-m2`
+
+### GLM Series
+* `accounts/fireworks/models/glm-4p5`
+* `accounts/fireworks/models/glm-4p5-air`
+* `accounts/fireworks/models/glm-4p6`
+
+### GPT-OSS Series
+* `accounts/fireworks/models/gpt-oss-20b`
+* `accounts/fireworks/models/gpt-oss-120b`
+
 ---
 
 ## Configuration in Roo Code
diff --git a/docs/providers/gemini.md b/docs/providers/gemini.md
@@ -46,6 +46,7 @@ Roo Code supports the main Gemini model families and automatically tracks Google
   - Roo Code defaults to a stable Pro model where your provider supports it (today this is a Gemini 2.5 Pro variant; future releases may point at newer Pro models)
 - **Gemini Flash family**
   - Fast, lower-cost models for everyday tasks and quick iterations
+- **Experimental:** `gemini-2.5-flash-lite-preview-09-2025` - Lightweight preview variant
 
 ### Aliases
 
diff --git a/docs/providers/mistral.md b/docs/providers/mistral.md
@@ -39,6 +39,9 @@ Roo Code supports the following Mistral models:
 |------------------------|-------------------------|------------------|--------|
 | codestral-latest      | 0.3                     | ✅               | ❌      |
 | mistral-large-latest  | 0.7                     | ✅               | ❌      |
+| magistral-medium-latest | 0.5                   | ✅               | ❌      |
+| devstral-medium-latest | 0.3                    | ✅               | ❌      |
+| mistral-medium-latest | 0.5                     | ✅               | ❌      |
 | ministral-8b-latest   | 0.3                     | ✅               | ❌      |
 | ministral-3b-latest   | 0.3                     | ✅               | ❌      |
 | mistral-small-latest  | 0.3                     | ✅               | ❌      |
diff --git a/docs/providers/openai.md b/docs/providers/openai.md
@@ -36,10 +36,17 @@ Roo Code supports accessing models directly through the official OpenAI API, inc
 
 ## Supported Models
 
-### GPT-5 Family (Latest)
-The GPT-5 models are OpenAI's most advanced, offering superior coding capabilities and agentic task performance:
+### GPT-5.1 Family (Latest)
+The GPT-5.1 models are OpenAI's most advanced, offering superior coding capabilities and agentic task performance:
 
-* **`gpt-5-2025-08-07`** (default) - Best model for coding and agentic tasks
+* **`gpt-5.1`** (default) - Best model for coding and agentic tasks
+* **`gpt-5.1-codex`** - Specialized coding variant
+* **`gpt-5.1-codex-mini`** - Fast, cost-efficient coding model
+
+### GPT-5 Family
+Earlier GPT-5 models:
+
+* **`gpt-5-2025-08-07`** - Best model for coding and agentic tasks
 * **`gpt-5-mini-2025-08-07`** - Faster, cost-efficient for well-defined tasks
 * **`gpt-5-nano-2025-08-07`** - Fastest, most cost-efficient option
 
@@ -88,6 +95,9 @@ Optimized GPT-4 models:
 * `gpt-4o` - Optimized GPT-4
 * `gpt-4o-mini` - Smaller optimized variant
 
+### Codex Models
+* `codex-mini-latest` - Cloud-based software engineering agent powered by codex-1, a version of o3 optimized for coding tasks
+
 Refer to the [OpenAI Models documentation](https://platform.openai.com/docs/models) for the most up-to-date list of models and capabilities.
 
 ---
@@ -99,7 +109,7 @@ Refer to the [OpenAI Models documentation](https://platform.openai.com/docs/mode
 1.  **Open Roo Code Settings:** Click the gear icon (<Codicon name="gear" />) in the Roo Code panel.
 2.  **Select Provider:** Choose "OpenAI" from the "API Provider" dropdown.
 3.  **Enter API Key:** Paste your OpenAI API key into the "OpenAI API Key" field.
-4.  **Select Model:** Choose your desired model from the "Model" dropdown (defaults to `gpt-5-2025-08-07`).
+4.  **Select Model:** Choose your desired model from the "Model" dropdown (defaults to `gpt-5.1`).
 5.  **(Optional) Base URL:** If you need to use a custom base URL, enter the URL. Most people won't need to adjust this.
 
 ---
diff --git a/docs/providers/sambanova.md b/docs/providers/sambanova.md
@@ -32,8 +32,10 @@ To use SambaNova with Roo Code, you'll need an API key from the [SambaCloud](htt
 Roo Code will attempt to fetch the list of available models from the SambaNova API. Common models available via SambaCloud include:
 
 *   `DeepSeek-R1`
+*   `DeepSeek-V3.1`
 *   `DeepSeek-V3-0324`
 *   `DeepSeek-R1-Distill-Llama-70B`
+*   `gpt-oss-120b`
 *   `Meta-Llama-3.3-70B-Instruct`
 *   `Meta-Llama-3.1-8B-Instruct`
 *   `Llama-4-Maverick-17B-128E-Instruct`
diff --git a/docs/providers/vertex.md b/docs/providers/vertex.md
@@ -39,7 +39,7 @@ Roo Code supports accessing models through Google Cloud Platform's Vertex AI, a
 Roo Code supports the following models through Vertex AI:
 
 ### Default Model
-If no model is specified, Roo Code defaults to `claude-sonnet-4@20250514`.
+If no model is specified, Roo Code defaults to `claude-sonnet-4-5@20250929`.
 
 ### Google Gemini Models
 
@@ -68,9 +68,10 @@ Some Gemini models provide dedicated reasoning or "thinking" tokens:
 Refer to the [Google Cloud Vertex AI models documentation](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models) for up-to-date Gemini model IDs and capabilities.
 
 ### Anthropic Claude Models
+*   `claude-sonnet-4-5@20250929` (Default)
 *   `claude-opus-4-1@20250805`
 *   `claude-opus-4@20250514`
-*   `claude-sonnet-4@20250514` (Default)
+*   `claude-sonnet-4@20250514`
 *   `claude-3-7-sonnet@20250219`
 *   `claude-3-7-sonnet@20250219:thinking` - With reasoning support
 *   `claude-3-5-sonnet-v2@20241022`
@@ -80,7 +81,18 @@ Refer to the [Google Cloud Vertex AI models documentation](https://cloud.google.
 *   `claude-3-haiku@20240307`
 
 ### Other Models
-*   `llama-4-maverick-17b-128e-instruct-maas` - Meta Llama 4 model with 128K context window
+
+#### MAAS (Model as a Service) Models
+*   `deepseek-r1-0528-maas`
+*   `deepseek-v3.1-maas`
+*   `gpt-oss-120b-maas`
+*   `gpt-oss-20b-maas`
+*   `qwen3-coder-480b-a35b-instruct-maas`
+*   `qwen3-235b-a22b-instruct-2507-maas`
+*   `llama-4-maverick-17b-128e-instruct-maas`
+
+#### Gemini Models
+*   `gemini-2.5-flash-lite-preview-06-17`
 
 Refer to the [Google Cloud documentation on Vertex AI Models](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models) for the most up-to-date list of available models and their IDs.
 
diff --git a/docs/providers/xai.md b/docs/providers/xai.md
@@ -40,6 +40,9 @@ Roo Code supports the following xAI Grok models:
 
 ### Current Models
 * `grok-code-fast-1` (Default) - xAI's Grok Code Fast model with 262K context window and prompt caching, optimized for reasoning and coding tasks
+* `grok-code-fast-1:2m` - Extended 2M context window variant
+* `grok-4.1-fast` - Latest Grok 4.1 fast model with advanced capabilities
+* `grok-4.1-fast:2m` - Grok 4.1 fast with 2M context window
 * `grok-4` - xAI's Grok-4 model with 262K context window, image support, and prompt caching
 * `grok-3` - xAI's Grok-3 model with 128K context window and prompt caching
 * `grok-3-fast` - xAI's Grok-3 fast model with 128K context window and prompt caching
@@ -69,11 +72,12 @@ Grok 3 Mini models feature specialized reasoning capabilities, allowing them to
 
 ### Reasoning-Enabled Models
 
-Reasoning is only supported by:
+Several Grok models have reasoning capabilities. However, only the Grok 3 Mini models support configurable reasoning effort control:
+
 * `grok-3-mini` - Supports reasoning effort control
 * `grok-3-mini-fast` - Supports reasoning effort control
 
-The standard Grok 3 models (`grok-3` and `grok-3-fast`) and Grok 4 do not support reasoning effort parameters.
+Other models (`grok-code-fast-1`, `grok-4.1-fast`, `grok-4`, `grok-3`, `grok-3-fast`) are reasoning-capable but don't expose the `reasoning_effort` parameter.
 
 ### Controlling Reasoning Effort
 
