Supported LLM Models
This document lists all supported large language models (LLMs) across providers, as defined in the codebase.
OpenAI
| Model Name |
Context Window |
Max Response |
Thinking Supported |
gpt-3.5-turbo |
16,385 |
4,096 |
No |
gpt-4.1 |
1,047,576 |
32,768 |
No |
gpt-4.1-mini |
1,047,576 |
32,768 |
No |
gpt-4.1-nano |
1,047,576 |
32,768 |
No |
gpt-4-turbo |
128,000 |
N/A |
No |
gpt-4o |
128,000 |
4,096 |
No |
gpt-4o-mini |
128,000 |
16,384 |
No |
o3-mini |
200,000 |
100,000 |
Yes |
o3 |
200,000 |
100,000 |
Yes |
o4-mini |
200,000 |
100,000 |
Yes |
gpt-5 |
200,000 |
100,000 |
Yes |
gpt-5-mini |
200,000 |
100,000 |
Yes |
gpt-5-nano |
200,000 |
100,000 |
Yes |
Note: gpt-4-turbo-alt is a duplicate entry for internal use and should not be selected by users.
Anthropic
| Model Name |
Max Response |
Thinking Supported |
claude-opus-4-20250514 |
32,000 |
No |
claude-sonnet-4-20250514 |
64,000 |
No |
claude-3-7-sonnet-20250219 |
64,000 |
No |
claude-3-5-haiku-20241022 |
8,192 |
No |
claude-3-5-sonnet-20241022 |
8,192 |
No |
claude-3-haiku-20240307 |
4,096 |
No |
Google Gemini
| Model Name |
Max Response |
Max COT |
Thinking Supported |
gemini-2.5-flash |
8,192 |
24,576 |
Yes |
gemini-2.5-pro |
65,536 |
196,608 |
Yes |
gemini-2.5-flash-lite-preview-06-17 |
64,000 |
192,000 |
Yes |
Mistral
| Model Name |
Context Window |
Max Input |
Max Response |
Thinking Supported |
codestral-latest |
256,000 |
250,000 |
4,096 |
No |
codestral-2405 |
256,000 |
250,000 |
4,096 |
No |
mistral-small-latest |
32,000 |
28,000 |
4,096 |
No |
mistral-medium-latest |
32,000 |
28,000 |
4,096 |
No |
mistral-large-latest |
128,000 |
120,000 |
4,096 |
No |
devstral-small-latest |
128,000 |
120,000 |
4,096 |
No |
devstral-medium-latest |
128,000 |
120,000 |
4,096 |
No |
Cerebras
| Model Name |
Max Input |
Max Response |
Description |
qwen-3-32b |
128,000 |
16,384 |
Qwen 3 32B model for general instruction following |
qwen-3-235b-a22b-instruct-2507 |
128,000 |
16,384 |
Qwen 3 235B A22B instruction-tuned model (preview) |
qwen-3-235b-a22b-thinking-2507 |
128,000 |
16,384 |
Qwen 3 235B A22B thinking model for reasoning tasks (preview) |
qwen-3-coder-480b |
128,000 |
16,384 |
Qwen 3 Coder 480B model for programming tasks (preview) |
gpt-oss-120b |
128,000 |
16,384 |
GPT-OSS 120B open-source model (preview) |
Z.AI
| Model Name |
Context Window |
Max Input |
Max COT |
Max Response |
Thinking Supported |
Supports Tools |
Supports Images |
glm-4.5 |
128,000 |
128,000 |
4,096 |
4,096 |
Yes |
Yes |
Yes |
glm-4.5-air |
128,000 |
128,000 |
4,096 |
4,096 |
Yes |
Yes |
Yes |
glm-4.6 |
128,000 |
128,000 |
4,096 |
4,096 |
Yes |
Yes |
Yes |
Alibaba Qwen
| Model Name |
Context Window |
Max Response |
Category |
qwen-turbo |
1,008,192 |
8,192 |
Alibaba Qwen Turbo Model (OpenAI-compatible) |
qwen-plus |
131,072 |
8,192 |
Alibaba Qwen Plus Model (OpenAI-compatible) |
qwen-flash |
1,000,000 |
8,192 |
Alibaba Qwen Flash Model (OpenAI-compatible) |
qwen-max |
32,768 |
8,192 |
Alibaba Qwen Max Model (OpenAI-compatible) |
qwen3-coder-plus |
1,048,576 |
65,536 |
Alibaba Qwen3 Coder Plus Model (OpenAI-compatible) |
qwen3-coder-480b-a35b-instruct |
262,144 |
65,536 |
Alibaba Qwen3 Coder 480B A35B Instruct Model (OpenAI-compatible) |
qwen3-235b-a22b-thinking-2507 |
131,072 |
32,768 |
Alibaba Qwen3 235B A22B Thinking Model (OpenAI-compatible) |
qwen3-235b-a22b-instruct-2507 |
129,024 |
32,768 |
Alibaba Qwen3 235B A22B Instruct Model (OpenAI-compatible) |
qwen3-30b-a3b-thinking-2507 |
126,976 |
32,768 |
Alibaba Qwen3 30B A3B Thinking Model (OpenAI-compatible) |
qwen3-30b-a3b-instruct-2507 |
129,024 |
32,768 |
Alibaba Qwen3 30B A3B Instruct Model (OpenAI-compatible) |
qwen3-next-80b-a3b-instruct |
262,144 |
65,536 |
Alibaba Qwen3-Max Preview (256K) - 80B A3B Instruct Model (OpenAI-compatible) |
qwen3-next-80b-a3b-thinking |
262,144 |
65,536 |
Alibaba Qwen3-Max Preview (256K) - 80B A3B Thinking Model (OpenAI-compatible) |
qwen3-max |
262,144 |
65,536 |
Alibaba Qwen3-Max Preview (256K) - Standard Model (OpenAI-compatible) |
DeepSeek
| Model Name |
Context Window |
Max Tokens |
Description |
deepseek-chat |
131,072 |
4,096 |
DeepSeek Chat Model (OpenAI-compatible) |
deepseek-reasoner |
131,072 |
32,768 |
DeepSeek Reasoner Model (OpenAI-compatible) |
Moonshot
| Model Name |
Context Window |
Max Input |
Max Response |
kimi-k2-0711-preview |
128,000 |
100,000 |
4,096 |
kimi-k2-turbo-preview |
128,000 |
100,000 |
4,096 |
kimi-k2-0905-preview |
128,000 |
100,000 |
4,096 |
IBM WatsonX
| Model Name |
Context Window |
Max Input |
Max Response |
Max COT |
Thinking Supported |
openai/gpt-oss-120b |
128,000 |
128,000 |
4,096 |
4,096 |
Yes |
ibm/granite-3-8b-instruct |
128,000 |
128,000 |
4,096 |
4,096 |
No |
ibm/granite-3-3-8b-instruct |
128,000 |
128,000 |
4,096 |
4,096 |
No |
meta-llama/llama-3-1-70b-instruct |
128,000 |
128,000 |
4,096 |
4,096 |
No |
meta-llama/llama-3-3-70b-instruct |
128,000 |
128,000 |
4,096 |
4,096 |
No |
mistralai/mistral-large |
128,000 |
128,000 |
4,096 |
4,096 |
No |
mistralai/mistral-large-2407 |
128,000 |
128,000 |
4,096 |
4,096 |
No |
openai/gpt-oss-20b |
128,000 |
128,000 |
4,096 |
4,096 |
Yes |
Azure OpenAI
Azure OpenAI supports any deployment name configured by the user. The following are known model mappings:
| Model Name |
Mapped To |
Context Window |
Max Response |
azure_openai_deployment |
gpt-4o |
128,000 |
4,096 |
Note: Azure OpenAI deployments are user-defined. Use --provider azure_openai --model YOUR_DEPLOYMENT_NAME to use any valid deployment.
Last updated from source code: September 2025