Models
Claude Sonnet 4.6
Anthropic's best balance of intelligence and speed — the recommended default for most production applications.
Models
Gemini 2.5 Pro
Google's best model — unmatched at multimodal tasks and long-context analysis, with a 1M token context window.
Models
GPT-5
OpenAI's most capable model, leading on ARC-AGI and general reasoning benchmarks as of April 2026.