Runefile — models: Section Reference
The models: section in Runefile configures LLM providers, model mapping, fallback policy, token budget, and safety.
Fields
| Field | Type | Default | Description |
providers | string[] | [] | Provider IDs in order of preference |
model_mapping | object | {} | Logical name → physical model ID |
fallback_policy | enum | next_provider | next_provider / fail |
token_budget | number | 100000 | Max tokens (input + output) per request |
safety_policy | enum | standard | standard / strict / none |
Supported Providers
| Provider ID | Required Env Var | Notes |
anthropic | ANTHROPIC_API_KEY | Claude Haiku / Sonnet / Opus |
openai | OPENAI_API_KEY | GPT-4o, GPT-4-turbo, o1, etc. |
gemini | GEMINI_API_KEY | Default model: gemini-2.0-flash (override with GEMINI_MODEL) |
copilot | GITHUB_TOKEN or COPILOT_TOKEN | GitHub Copilot — OpenAI-compatible endpoint |
claude-code | CLAUDE_API_KEY | Anthropic via a separate API key alias |
Auto-detection order (when multiple keys are set): anthropic → claude-code → gemini → copilot → openai
Example
providers:
- anthropic
- gemini
- openai
model_mapping:
default: claude-sonnet-4-6
fast: claude-haiku-4-5-20251001
fallback: gpt-4o-mini
fallback_policy: next_provider
token_budget: 100000
safety_policy: standard
Common Model IDs
| Provider | Model IDs |
| Anthropic | claude-sonnet-4-6, claude-haiku-4-5-20251001, claude-opus-4-6 |
| OpenAI | gpt-4o, gpt-4o-mini, gpt-4-turbo, o1-preview |
| Gemini | gemini-2.0-flash, gemini-1.5-pro, gemini-2.0-pro |
| Copilot | gpt-4o |