Available Models
Below is a list of all available models. Models are categorized by type:- Frontier: Best coding agent models
- Research: Expensive and limited heavily
- General: Cheaper but less intelligent models
| ID | Name | Type | Limit |
|---|---|---|---|
claude-opus-4-20250514 | Claude Opus 4 | Research | 200K |
claude-opus-4-1-20250805 | Claude Opus 4.1 | Research | 200K |
o3-pro | OpenAI-o3-pro | Research | 200K |
claude-haiku-4-5-20251001 | Claude Haiku 4.5 | Frontier | 200K |
claude-sonnet-4-20250514 | Claude Sonnet 4 | Frontier | 200K |
claude-sonnet-4-1m | Claude Sonnet 4 (1M) | Frontier | 1.0M |
claude-sonnet-4-5-20250929 | Claude Sonnet 4.5 | Frontier | 200K |
claude-sonnet-4-5-1m | Claude Sonnet 4.5 (1M) | Frontier | 1.0M |
gpt-5 | GPT-5 | Frontier | 400K |
gpt-5-high | GPT-5 (High Reasoning) | Frontier | 400K |
gpt-5-high-fast | GPT-5 (High Reasoning, Priority) | Frontier | 400K |
gpt-5-low | GPT-5 (Low Reasoning) | Frontier | 400K |
gpt-5-low-fast | GPT-5 (Low Reasoning, Priority) | Frontier | 400K |
gpt-5-fast | GPT-5 (Priority) | Frontier | 400K |
gpt-5-codex | GPT-5 Codex | Frontier | 400K |
gpt-5-codex-high | GPT-5 Codex (High Reasoning) | Frontier | 400K |
gpt-5-codex-low | GPT-5 Codex (Low Reasoning) | Frontier | 400K |
claude-3.5-sonnet | Claude-3.5 Sonnet | General | 200K |
claude-3.7-sonnet | Claude-3.7 Sonnet | General | 200K |
deepseekr1 | Deepseek r1 | General | 164K |
deepseekv3 | Deepseek v3 | General | 128K |
gpt-4.1 | GPT-4.1 | General | 1.0M |
gpt-4o | GPT-4o | General | 128K |
gpt-5-mini | GPT-5-mini | General | 400K |
gpt-oss-120b | GPT-OSS-120b | General | 131K |
gemini-2.5-pro | Gemini 2.5 Pro | General | 1.0M |
grok-4 | Grok 4 | General | 256K |
grok-4-fast | Grok 4 Fast | General | 2.0M |
grok-code-fast-1 | Grok Code Fast 1 | General | 256K |
grok-3 | Grok-3 | General | 131K |
o3 | OpenAI-o3 | General | 200K |
o3-mini | OpenAI-o3-mini | General | 200K |
o4-mini | OpenAI-o4-mini | General | 200K |
qwen-3-coder-480b | Qwen-3-Coder-480b | General | 131K |
Models Picker
You can see what model you are currently using by clicking the dropdown at the bottom left of the chat interface:
Models can misreport their name/version at runtime. Rely on the model ID you selected in Firebender, not the model’s self-identification response.
Example: Deepseek v3 misidentifies itself as gpt-4
Context Windows
A context window is the maximum span of tokens (text and code) a model can consider at once. The more prompts, files, and responses in a session, the more context is consumed. If you run into context window limit, its recommended to summarize or start a new chat.Avoiding Window Limits
- prune chats frequently if the coding agent does something wrong (instead of telling the agent to undo its changes, restart the prompt with approaches to avoid)
- scope down changes that can be easily verifiable (e.g. running tests, reviewing UI change, etc.)
Default Mode
Default Mode automatically selects the best model for each task based on request complexity and availability. When a rate limit is hit, Firebender automatically switches to the next best model.Controls
You can specify models using one of the following methods:- Deep Links:
jetbrains://idea/firebender/chat?model=claude-sonnet-4-20250514 - Commands: Configure in firebender.json with
"model": "claude-sonnet-4-20250514" - Plugin Interface: Select from the model dropdown in the Firebender chat interface