AI Models
Choose and configure AI models for code review, chat, and auto-fix.
GitSniff supports 20+ AI models from multiple providers through OpenRouter. You can configure which models are used for different scenarios.
Browse all available models — View the complete list of supported models with pricing, context windows, and capabilities on our Models page.
Model tiers
Models are organized into four tiers based on capability and cost:
| Tier | Characteristics | Best for |
|---|---|---|
| Flash | Fastest, lowest cost | Quick reviews, small PRs, simple changes |
| Fast | Good speed, moderate cost | Everyday reviews, standard PRs |
| Balanced | Good capability, moderate cost | Complex features, medium PRs |
| Powerful | Highest capability, highest cost | Critical PRs, security reviews, architecture changes |
Understanding model specifications
When choosing a model, consider these key specifications visible on our Models page:
Context window
The context window determines how much text a model can process in a single request. This includes both the code being reviewed and the model's response.
- Why it matters for code review: Large PRs with many changed files require more context. If a PR exceeds the model's context window, GitSniff must split the analysis, which can reduce review quality.
- Typical ranges: Context windows range from 8K tokens (basic models) to 1M+ tokens (advanced models). For reference, 1K tokens ≈ 750 words of code.
- Recommendation: For large PRs or monorepo reviews, prefer models with larger context windows. For small, focused PRs, smaller context windows are sufficient and often faster.
Seed parameter
Some models support a seed parameter for reproducible outputs.
- What it does: When a seed is set, the model attempts to produce the same output given the same input. This makes reviews more deterministic.
- Use cases: Useful when you want consistent review behavior across similar PRs, or when debugging why a review produced certain results.
- Availability: Not all models support seeds. Check the Models page to see which models have seed support.
Model selection modes
Auto mode (recommended)
GitSniff automatically selects the best model based on PR size:
- Small PRs — Uses the model configured for small PRs (default: Flash tier)
- Medium PRs — Uses the model configured for medium PRs (default: Balanced tier)
- Large PRs — Uses the model configured for large PRs (default: Powerful tier)
You can customize which specific model is used for each size category.
Manual mode
Select a specific model that's used for all PRs regardless of size. Useful when you want consistent results from a particular model.
Configuration
Configure model preferences in your settings:
- Organization level: Dashboard > Settings > AI Models
- Repository level: Dashboard > Repository > Settings > AI Models