Models / Best of /coding
Updated October 2026 · ranked from live data
Best models for coding
Ranked by our coding rating where we have one — how close each model sits to the board leader — then by recency among models positioned for code. Right now the top pick is Claude Opus 5.5 (5/5 coding (Arena Coding, 3 of 168 via High)). Every row links to full specs, hardware verdicts and provider pricing.
1Claude Opus 5.5Anthropic5/5 coding (Arena Coding, 3 of 168 via High)—$4.00 / $20.00per 1M · via Anthropic2Claude Fable 5Anthropic5/5 coding (Arena Coding, 1 of 168)—$10.00 / $50.00per 1M · via Anthropic3Claude Opus 4.7Anthropic5/5 coding (Arena Coding, 4 of 168)—$5.00 / $25.00per 1M · via Anthropic4Claude Opus 4.6Anthropic5/5 coding (Arena Coding, 2 of 168)—$5.00 / $25.00per 1M · via Anthropic5DeepSeek V4.1 FlashDeepSeek4.5/5 coding (Arena Coding, 11 of 168 via Max)763B$0.12 / $0.40per 1M · via DekaLLM6GPT-6 AstraOpenAI4.5/5 coding (Arena Coding, 5 of 168 via Max)—$10.00 / $50.00per 1M · via Microsoft Azure AI7Gemini 3.8 FlashGoogle4.5/5 coding (Arena Coding, 12 of 168 via High)—$0.75 / $3.75per 1M · via Google AI8Muse Spark 1.3Meta4.5/5 coding (Arena Coding, 7 of 168 via Max)—$1.25 / $4.25per 1M · via Meta9Claude Fable 5.1Anthropic4.5/5 coding (Arena Coding, 15 of 168 via Max)—$10.00 / $50.00per 1M · via Anthropic10Muse Spark 1.2Meta4.5/5 coding (Arena Coding, 8 of 168 via xHigh)—$1.25 / $4.25per 1M · via Meta
