Models
Pick the right model for the job. Names map 1:1 to what you pass in your requests.
How to choose
Building an agent?
Reach for claude-sonnet-4 — strongest at tools and multi-step reasoning.
High volume?
gpt-4.1-mini is fast and economical — the best value per token.
Huge documents?
gemini-2.5-pro handles very long context and multimodal inputs.
claude-sonnet-4
Coding and agents
High cost
3.2s avg
99.1%
Long context
StreamingTools
Healthy
gpt-4.1-mini
Everyday tasks
Low cost
1.1s avg
99.8%
Standard context
StreamingVisionTools
Healthy
gemini-2.5-pro
Long context
Medium cost
4.7s avg
97.9%
Very long context
Multimodal
Watch