ModelPrimary useLatency p50Cost / reviewReliabilityStructEnabled
google/gemini-2.5-flash1M
—540ms$0.002899.6%✓mistral/codestral-2128k
—660ms$0.002998.0%✓xai/grok-4-fast256k
—690ms$0.004297.8%✓anthropic/claude-haiku-4-5200k
Correctness · Tests740ms$0.006199.7%✓meta/llama-4-maverick256k
—820ms$0.002497.1%✓openai/gpt-5-mini400k
Fallback · Correctness980ms$0.007299.5%✓deepseek/deepseek-v3.2128k
Cost-tier fallback1100ms$0.000996.2%—google/gemini-2.5-pro1M
Frontend UX1320ms$0.018098.4%✓qwen/qwen3-coder-480b256k
Tests fallback1480ms$0.003896.8%✓anthropic/claude-sonnet-4-5200k
Security · Architecture1840ms$0.024099.4%✓openai/gpt-5400k
Architecture2210ms$0.041098.9%✓anthropic/claude-opus-4-1200k
Critical paths3120ms$0.087099.2%✓