Skip to content

Commit d4d542c

Browse files
SannidhyaSannidhya
Sannidhya
authored and
Sannidhya
committed
docs: simplify provider model lists by linking to authoritative sources
Replace exhaustive model ID lists with links to provider documentation. This eliminates maintenance burden from constantly updating model lists and ensures users always see current, accurate information. Changes: - Remove detailed model catalogs from 28 provider docs - Add links to official provider model documentation - Preserve Roo Code-specific guidance (2-4 recommended models) - Keep special integration features (reasoning effort, troubleshooting, etc.) Benefits: - No stale documentation from daily model changes - Single source of truth (provider APIs) - Reduced user confusion - Lower maintenance overhead
1 parent c9224d7 commit d4d542c

29 files changed

+255
-470
lines changed

‎docs/providers/anthropic.md‎

Lines changed: 10 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -32,23 +32,16 @@ Anthropic is an AI safety and research company that builds reliable, interpretab
3232

3333
---
3434

35-
## Supported Models
36-
37-
Roo Code supports the following Anthropic Claude models:
38-
39-
* `claude-sonnet-4-5` (Latest, Recommended)
40-
* `claude-opus-4-5-20251101`
41-
* `claude-opus-4-1-20250805`
42-
* `claude-opus-4-20250514`
43-
* `claude-sonnet-4-20250514`
44-
* `claude-3-7-sonnet-20250219`
45-
* `claude-3-7-sonnet-20250219:thinking` (Extended Thinking variant)
46-
* `claude-3-5-sonnet-20241022`
47-
* `claude-3-5-haiku-20241022`
48-
* `claude-3-opus-20240229`
49-
* `claude-3-haiku-20240307`
50-
51-
See [Anthropic's Model Documentation](https://docs.anthropic.com/en/docs/about-claude/models) for more details on each model's capabilities.
35+
## Available Models
36+
37+
Roo Code supports all Claude models available through Anthropic's API.
38+
39+
For the complete, up-to-date model list and capabilities, see [Anthropic's model documentation](https://docs.anthropic.com/en/docs/about-claude/models).
40+
41+
**Recommended for Roo Code:**
42+
- **Sonnet models** - Best balance of performance and cost for most coding tasks (default)
43+
- **Opus models** - Better for complex reasoning and large-scale refactoring
44+
- **Haiku models** - Faster and more cost-effective for simpler tasks
5245

5346
---
5447

‎docs/providers/bedrock.md‎

Lines changed: 14 additions & 47 deletions
Original file line numberDiff line numberDiff line change
@@ -47,53 +47,20 @@ You have two main options for configuring AWS credentials:
4747

4848
---
4949

50-
## Supported Models
51-
52-
Roo Code supports the following models through Bedrock (based on source code):
53-
54-
* **Amazon:**
55-
* `amazon.nova-pro-v1:0`
56-
* `amazon.nova-pro-latency-optimized-v1:0`
57-
* `amazon.nova-lite-v1:0`
58-
* `amazon.nova-micro-v1:0`
59-
* `amazon.titan-text-lite-v1:0`
60-
* `amazon.titan-text-express-v1:0`
61-
* `amazon.titan-text-embeddings-v1:0`
62-
* `amazon.titan-text-embeddings-v2:0`
63-
* **Anthropic:**
64-
* `anthropic.claude-sonnet-4-5-20250929-v1:0` (Default)
65-
* `anthropic.claude-opus-4.1-20250514-v1:0`
66-
* `anthropic.claude-opus-4-20250514-v1:0`
67-
* `anthropic.claude-sonnet-4-20250514-v1:0`
68-
* `anthropic.claude-3-7-sonnet-20250219-v1:0`
69-
* `anthropic.claude-3-5-sonnet-20241022-v2:0`
70-
* `anthropic.claude-3-5-haiku-20241022-v1:0`
71-
* `anthropic.claude-3-5-sonnet-20240620-v1:0`
72-
* `anthropic.claude-3-opus-20240229-v1:0`
73-
* `anthropic.claude-3-sonnet-20240229-v1:0`
74-
* `anthropic.claude-3-haiku-20240307-v1:0`
75-
* `anthropic.claude-2-1-v1:0`
76-
* `anthropic.claude-2-0-v1:0`
77-
* `anthropic.claude-instant-v1:0`
78-
* **DeepSeek:**
79-
* `deepseek.r1-v1:0`
80-
* **Meta:**
81-
* `meta.llama3-3-70b-instruct-v1:0`
82-
* `meta.llama3-2-90b-instruct-v1:0`
83-
* `meta.llama3-2-11b-instruct-v1:0`
84-
* `meta.llama3-2-3b-instruct-v1:0`
85-
* `meta.llama3-2-1b-instruct-v1:0`
86-
* `meta.llama3-1-405b-instruct-v1:0`
87-
* `meta.llama3-1-70b-instruct-v1:0`
88-
* `meta.llama3-1-70b-instruct-latency-optimized-v1:0`
89-
* `meta.llama3-1-8b-instruct-v1:0`
90-
* `meta.llama3-70b-instruct-v1:0`
91-
* `meta.llama3-8b-instruct-v1:0`
92-
* **OpenAI:**
93-
* `openai.gpt-oss-20b-1:0`
94-
* `openai.gpt-oss-120b-1:0`
95-
96-
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.
50+
## Available Models
51+
52+
Roo Code supports all foundation models available through Amazon Bedrock.
53+
54+
For the complete, up-to-date model list with IDs and capabilities, see [AWS Bedrock's supported models documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html).
55+
56+
**Important:** Use the *model ID* (e.g., `anthropic.claude-sonnet-4-5-20250929-v1:0`) when configuring Roo Code, not the model name.
57+
58+
**Recommended for Roo Code:**
59+
- **Claude Sonnet models** - Best balance for most coding tasks (default: `anthropic.claude-sonnet-4-5-20250929-v1:0`)
60+
- **Amazon Nova models** - Better for AWS-integrated workflows
61+
- **Meta Llama models** - Good for open-source requirements
62+
63+
**Note:** Model availability varies by AWS region. Request access to specific models through the Bedrock console before use.
9764
9865
---
9966

‎docs/providers/cerebras.md‎

Lines changed: 14 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -41,9 +41,20 @@ Cerebras AI specializes in extremely fast inference speeds (up to 2600 tokens/se
4141

4242
---
4343

44+
## Available Models
45+
46+
Roo Code automatically fetches all available models from Cerebras AI's API.
47+
48+
For the complete, up-to-date model list and pricing, see [Cerebras Cloud](https://cloud.cerebras. ai?utm_source=roocode).
49+
50+
**Key advantages:**
51+
- Ultra-fast inference (up to 2600 tokens/second)
52+
- Free tier available with rate limits
53+
- Context windows: 64K-128K tokens
54+
55+
---
56+
4457
## Tips and Notes
4558

4659
* **Performance:** Cerebras specializes in extremely fast inference speeds, making it ideal for real-time coding assistance.
47-
* **Free Tier:** The `qwen-3-coder-480b-free` model provides access to high-performance inference at no cost with rate limits.
48-
* **Context Windows:** Models support context windows ranging from 64K to 128K tokens.
49-
* **Pricing:** Refer to the [Cerebras Cloud](https://cloud.cerebras.ai?utm_source=roocode) dashboard for the latest pricing information.
60+
* **Pricing:** Check the [Cerebras Cloud](https://cloud.cerebras.ai?utm_source=roocode) dashboard for current pricing and free tier details.

‎docs/providers/chutes.md‎

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -27,11 +27,13 @@ To use Chutes AI with Roo Code, obtain an API key from the [Chutes AI platform](
2727

2828
---
2929

30-
## Supported Models
30+
## Available Models
3131

32-
Roo Code will attempt to fetch the list of available models from the Chutes AI API. The specific models available will depend on Chutes AI's current offerings.
32+
Roo Code automatically fetches all available models from Chutes AI's API.
3333

34-
Always refer to the official Chutes AI documentation or your dashboard for the most up-to-date list of supported models.
34+
For the complete, up-to-date model list, see [Chutes AI's platform](https://chutes.ai/) or your account dashboard.
35+
36+
**Key advantage:** Free API access to multiple LLMs for experimentation and development.
3537

3638
---
3739

‎docs/providers/claude-code.md‎

Lines changed: 7 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -94,18 +94,16 @@ export CLAUDE_CODE_MAX_OUTPUT_TOKENS=32768 # Set to 32k tokens
9494

9595
---
9696

97-
## Supported Models
97+
## Available Models
9898

99-
The Claude Code provider supports these Claude models:
99+
The Claude Code provider supports all Claude models available through the official CLI.
100100

101-
- **Claude Opus 4.1** (Most capable)
102-
- **Claude Opus 4**
103-
- **Claude Sonnet 4** (Latest, recommended)
104-
- **Claude 3.7 Sonnet**
105-
- **Claude 3.5 Sonnet**
106-
- **Claude 3.5 Haiku** (Fast responses)
101+
Model availability depends on your Claude CLI subscription and plan. See [Anthropic's CLI documentation](https://docs.anthropic.com/en/docs/claude-code/setup) for details.
107102

108-
The specific models available depend on your Claude CLI subscription and plan.
103+
**Recommended:**
104+
- **Sonnet models** - Best balance for most coding tasks (latest recommended)
105+
- **Opus models** - Better for complex reasoning
106+
- **Haiku models** - Faster responses when speed matters
109107

110108

111109
---

‎docs/providers/deepinfra.md‎

Lines changed: 8 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -31,20 +31,18 @@ DeepInfra provides cost-effective access to high-performance open-source models
3131

3232
---
3333

34-
## Supported Models
34+
## Available Models
3535

36-
Roo Code dynamically fetches available models from DeepInfra's API. The default model is:
36+
Roo Code automatically fetches all available models from DeepInfra's API.
3737

38-
* `Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo` (256K context, optimized for coding)
38+
For the complete, up-to-date model catalog, see [DeepInfra's models page](https://deepinfra.com/models).
3939

40-
Common models available include:
40+
**Recommended for Roo Code:**
41+
- **Qwen Coder models** - Best for programming tasks with large context windows (default: `Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo`)
42+
- **Vision-capable models** - Better when you need image understanding
43+
- **Reasoning models** - Best for complex problem-solving tasks
4144

42-
* **Coding Models:** Qwen Coder series, specialized for programming tasks
43-
* **General Models:** Llama 3.1, Mixtral, and other open-source models
44-
* **Vision Models:** Models with image understanding capabilities
45-
* **Reasoning Models:** Models with advanced reasoning support
46-
47-
Browse the full catalog at [deepinfra.com/models](https://deepinfra.com/models).
45+
**Key features:** Prompt caching support, low latency with global edge locations, competitive pricing.
4846

4947
---
5048

‎docs/providers/deepseek.md‎

Lines changed: 7 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -30,12 +30,15 @@ Roo Code supports accessing models through the DeepSeek API, including `deepseek
3030

3131
---
3232

33-
## Supported Models
33+
## Available Models
3434

35-
Roo Code supports the following DeepSeek models:
35+
Roo Code supports all models available through the DeepSeek API.
3636

37-
* `deepseek-chat` (Recommended for coding tasks)
38-
* `deepseek-reasoner` (Recommended for reasoning tasks)
37+
For the complete, up-to-date model list, see [DeepSeek's API documentation](https://api-docs.deepseek.com/quick_start/pricing).
38+
39+
**Recommended for Roo Code:**
40+
- **`deepseek-chat`** - Best for general coding tasks
41+
- **`deepseek-reasoner`** - Better for complex reasoning and problem-solving
3942

4043
---
4144

‎docs/providers/doubao.md‎

Lines changed: 11 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -30,20 +30,22 @@ Doubao is ByteDance's Chinese AI service, offering competitive language models f
3030

3131
---
3232

33-
## Supported Models
33+
## Available Models
3434

35-
Roo Code supports the following Doubao models:
35+
Roo Code supports all Doubao models available through ByteDance's Volcano Engine API.
3636

37-
* `doubao-seed-1-6-250615` (Default) - General purpose
38-
* `doubao-seed-1-6-thinking-250715` - Enhanced reasoning
39-
* `doubao-seed-1-6-flash-250715` - Speed optimized
37+
For the complete, up-to-date model list, see [Volcano Engine's AI model service](https://www.volcengine.com/).
4038

41-
All models support:
42-
- 128,000 token context window
43-
- 32,768 max output tokens
44-
- Image inputs
39+
**Model features:**
40+
- 128K context window
41+
- Image input support
4542
- Prompt caching with 80% discount on cached reads
4643

44+
**Recommended:**
45+
- General purpose: Standard models for everyday tasks
46+
- Thinking models: Better for enhanced reasoning
47+
- Flash models: Faster for speed-optimized workflows
48+
4749
---
4850

4951
## Configuration in Roo Code

‎docs/providers/featherless.md‎

Lines changed: 9 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -32,26 +32,20 @@ Featherless AI provides access to high-performance open-source models including
3232

3333
---
3434

35-
## Supported Models
35+
## Available Models
3636

37-
Roo Code supports the following Featherless models:
37+
Roo Code automatically fetches all available models from Featherless AI's API.
3838

39-
* `deepseek-ai/DeepSeek-R1-0528` (Default) - DeepSeek R1 reasoning model with `<think>` tag support
40-
* `deepseek-ai/DeepSeek-V3-0324` - DeepSeek V3 model
41-
* `moonshotai/Kimi-K2-Instruct` - Kimi K2 instruction-following model
42-
* `openai/gpt-oss-120b` - GPT-OSS 120B parameter model
43-
* `Qwen/Qwen3-Coder-480B-A35B-Instruct` - Qwen3 specialized coding model
39+
For the complete, up-to-date model list, see [Featherless AI](https://featherless.ai).
4440

45-
### Model Capabilities
41+
**All models are currently FREE** with no usage costs.
4642

47-
All models support:
48-
- **Context Window:** ~32,678 tokens
49-
- **Max Output:** 4,096 tokens
50-
- **Pricing:** Free (no cost for input/output tokens)
43+
**Recommended for Roo Code:**
44+
- **DeepSeek R1 models** - Best for complex reasoning with `<think>` tag support (default)
45+
- **Qwen3 Coder** - Better for specialized code generation tasks
46+
- **Kimi K2** - Good for balanced instruction-following
5147

52-
:::info
53-
**DeepSeek R1 Models:** The DeepSeek R1 models (like `DeepSeek-R1-0528`) include special reasoning capabilities with `<think>` tag support for step-by-step problem solving. These models automatically separate reasoning from regular output.
54-
:::
48+
**Note:** Most models have ~32K context window and 4K max output. No image support or prompt caching available.
5549

5650
---
5751

‎docs/providers/fireworks.md‎

Lines changed: 11 additions & 45 deletions
Original file line numberDiff line numberDiff line change
@@ -38,51 +38,17 @@ Fireworks AI is a high-performance platform for running AI models, offering fast
3838

3939
---
4040

41-
## Supported Models
42-
43-
Fireworks AI offers several high-performance models:
44-
45-
### Kimi K2
46-
* `accounts/fireworks/models/kimi-k2-instruct-0905` (Default)
47-
- 1 trillion total parameters with 32B activated params
48-
- 256K context window
49-
- Optimized for agentic capabilities
50-
- $0.60/M input, $2.50/M output
51-
52-
### Qwen3 Series
53-
* `accounts/fireworks/models/qwen3-235b-a22b-instruct-2507`
54-
- 256K context window
55-
- Competitive with closed-source models
56-
- $0.22/M input, $0.88/M output
57-
58-
* `accounts/fireworks/models/qwen3-coder-480b-a35b-instruct`
59-
- 256K context window
60-
- Specialized for coding tasks
61-
- $0.45/M input, $1.80/M output
62-
63-
### DeepSeek Series
64-
* `accounts/fireworks/models/deepseek-r1-0528`
65-
- 160K context window
66-
- Advanced reasoning with reduced hallucination
67-
- Function calling support
68-
- $3.00/M input, $8.00/M output
69-
70-
* `accounts/fireworks/models/deepseek-v3`
71-
- 128K context window
72-
- 671B total parameters with 37B activated
73-
- $0.90/M input, $0.90/M output
74-
75-
### MiniMax
76-
* `accounts/fireworks/models/minimax-m2`
77-
78-
### GLM Series
79-
* `accounts/fireworks/models/glm-4p5`
80-
* `accounts/fireworks/models/glm-4p5-air`
81-
* `accounts/fireworks/models/glm-4p6`
82-
83-
### GPT-OSS Series
84-
* `accounts/fireworks/models/gpt-oss-20b`
85-
* `accounts/fireworks/models/gpt-oss-120b`
41+
## Available Models
42+
43+
Roo Code supports all models available through Fireworks AI's platform.
44+
45+
For the complete, up-to-date model list and specifications, see [Fireworks AI's models page](https://fireworks.ai/models).
46+
47+
**Recommended for Roo Code:**
48+
- **Kimi K2** - Best for general-purpose coding with agentic capabilities (default)
49+
- **Qwen3 Coder** - Better for specialized code generation and debugging
50+
- **DeepSeek R1** - Best for complex reasoning and function calling tasks
51+
- **Qwen3 235B** - Most cost-effective for general development work
8652

8753
---
8854

0 commit comments

Comments
 (0)