Head to head
Two models, one table: published price per million tokens, context window, intelligence index and independent benchmarks. The winning cell on each row is marked, and every price links back to the provider page it was read from. Comparing more than two? The whole catalogue sorts on any column in the catalogue.
GPT-6 Astra (OpenAI) vs Claude Fable 5 (Anthropic)
Same blended price, to within 5%
frontier flagships
Claude Fable 5 (Anthropic) vs Gemini 3.8 Flash (Google)
Gemini 3.8 Flash is 13× cheaper per blended million
price performance
Grok 4.6 (xAI) vs GPT-6 Astra (OpenAI)
Grok 4.6 is 7.5× cheaper per blended million
frontier flagships
DeepSeek V4 Pro 0813 (DeepSeek) vs GPT-6 Astra (OpenAI)
DeepSeek V4 Pro 0813 is 23× cheaper per blended million
budget vs frontier
Kimi K3 (Moonshot AI) vs DeepSeek V4 Pro 0813 (DeepSeek)
DeepSeek V4 Pro 0813 is 6.8× cheaper per blended million
open weight flagships
Qwen3 Coder Next (Alibaba) vs Kimi K2.7 Code (Moonshot AI)
Qwen3 Coder Next is 2.8× cheaper per blended million
coding models
Claude Sonnet 5 (Anthropic) vs GPT-5.6 Sol (OpenAI)
Claude Sonnet 5 is 2.0× cheaper per blended million
mid tier workhorses
Prices as of 2026-10-23. Latency figures come from the weekly run described in the methodology; a row is left blank rather than filled with a number measured somewhere else. Looking for a pairing that is not here? The rankings put every priced model on the same scale.