Models / Best of /coding
Updated August 2026 · ranked from live data
Best models for coding
Ranked by measured coding benchmark percentile where we have it, then by recency among models positioned for code.Right now the top pick is Claude Fable 5 (5/5 coding (Arena Coding, 1 of 143)).Every row links to full specs, hardware verdicts and provider pricing.
1Claude Fable 5Anthropic5/5 coding (Arena Coding, 1 of 143)—$10.00 / $50.00per 1M · via Microsoft Azure AI2Claude Opus 4.7Anthropic5/5 coding (Arena Coding, 3 of 143)—$5.00 / $25.00per 1M · via Microsoft Azure AI3Claude Opus 4.6Anthropic5/5 coding (Arena Coding, 2 of 143)—$5.00 / $25.00per 1M · via Anthropic4Claude Opus 5Anthropic4.5/5 coding (Arena Coding, 4 of 143)—$5.00 / $25.00per 1M · via Anthropic5Muse Spark 1.1Meta4.5/5 coding (Arena Coding, 6 of 143)—$1.25 / $4.25per 1M · via Meta6Kimi K3Moonshot AI4.5/5 coding (Arena Coding, 5 of 143)2.8T / 104B$3.00 / $15.00per 1M · via Novita AI7Gemini 3.5 Flash LiteGoogle4/5 coding (Arena Coding, 28 of 143)—$0.30 / $2.50per 1M · via Google AI8Gemini 3.6 FlashGoogle4/5 coding (Arena Coding, 12 of 143)—$1.50 / $7.50per 1M · via Google AI9GPT-5.6 TerraOpenAI4/5 coding (Arena Coding, 14 of 143)—$1.00 / $6.00per 1M · via OpenRouter10GPT-5.6 SolOpenAI4/5 coding (Arena Coding, 8 of 143)—$5.00 / $30.00per 1M · via OpenRouter
