Skip to main content

Available Models

Access Claude and Gemini models via Google Vertex AI.

Claude Models

Model IDDescriptionBest ForSpeed
claude-opus-4-5-thinkingMost capable model with extended thinkingComplex tasks, architecture, planningSlower
claude-sonnet-4-5-thinkingFast and capable with extended thinkingDaily coding, quick tasksFast
claude-sonnet-4-5Standard Sonnet without extended thinkingGeneral use, fastest ClaudeFastest

Gemini Models

Model IDDescriptionBest For
gemini-3-flashVery fast responsesQuick edits, simple tasks, preserving quota
gemini-3-pro-highHigher quality GeminiAlternative when Claude quota is low
Gemini models are useful when you want to preserve Claude quota for complex tasks, or when you need very fast responses for simple operations.

{
  "ANTHROPIC_MODEL": "claude-opus-4-5-thinking",
  "ANTHROPIC_DEFAULT_OPUS_MODEL": "claude-opus-4-5-thinking",
  "ANTHROPIC_DEFAULT_SONNET_MODEL": "claude-sonnet-4-5-thinking",
  "ANTHROPIC_DEFAULT_HAIKU_MODEL": "gemini-3-flash"
}
Use /model model-name to switch during a session.

Model Selection Guide

TaskRecommended ModelWhy
Architecture decisionsclaude-opus-4-5-thinkingNeeds deep reasoning
Code generationclaude-sonnet-4-5-thinkingGood balance of speed/quality
Complex debuggingclaude-opus-4-5-thinkingExtended thinking helps
Documentationclaude-sonnet-4-5-thinkingFast enough for prose
Quick fixesgemini-3-flashVery fast, saves quota
Boilerplategemini-3-flashDoesn’t need Claude quality

Extended Thinking

Models with -thinking suffix use extended thinking mode:
  • Claude thinks through problems step-by-step before responding
  • Better for complex reasoning and multi-step tasks
  • Slightly slower but higher quality for difficult problems
If you don’t need extended thinking, use claude-sonnet-4-5 for faster responses.