Arena studied 105 models across 44 weekly Code Arena: WebDev boards since December 2025. Xiaomi's MiMo-V2.6-Flash now beats April's leader Claude Opus 4.7 at under a twentieth of the price, the top score has risen about 41 points a month since May (3x the winter pace), and 19 of the 25 price-performance frontier models under $1 per answer are Chinese.
Key Takeaways
- ✓Cadence: about 5 new models a month last winter, 12 to 14 a month from July to September, and 11 by October 23; OpenAI listed 12, Anthropic, Alibaba and Google 11 each
- ✓Top score: April's best, Claude Opus 4.7, scores 1558 on today's board versus 1807 for today's leader Claude Opus 5.5; a new top score now lasts about three weeks instead of two months
- ✓Price: between 5 cents and $1 per answer, today's best has an estimated 65-80% win rate against April's best at the same price; Opus 5.5 (Max) has about 81% but costs about $10 per answer
- ✓Volatility: nearly half the top 10 labs change rank each week, and the #10 lab cut its gap to #1 by 38% (282 to 176 points)
- ✓Regional split: 13 of the 15 frontier models above $1 are American; 17 of 21 Chinese frontier models have open weights versus 2 of 19 American ones
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and pain points
In late October, new models from Meta, MiniMax, NVIDIA and Moonshot hit Code Arena within days, and 19 models joined the board in just over four weeks from late September. The question for developers is whether this pace is a blip or the norm, and how to read leaderboards when picking a model. Arena researcher Dawid Galarowicz answered it in a long X article reviewing the full history of the Code Arena: WebDev board.
Methodology
The study covers 44 weekly boards from December 1, 2025 to October 19, 2026 and the 105 newly listed models, counting effort settings such as High and Max once. Scores come only from anonymous community battles. Cost per answer is the monthly median provider-billed cost in real battles, available from April 2026. Win rates are estimated from score gaps (every 400 points multiplies the odds by 10), not counted from head-to-head battles.
Benchmarks and data
April's leader, Claude Opus 4.7, scores 1558 on today's board, against 1807 for today's leader, Claude Opus 5.5. Xiaomi's MiMo-V2.6-Flash now beats Opus 4.7 at under a twentieth of the price. Since May the top score has risen about 41 points a month, versus about 13 over the winter. Anthropic held #1 on 39 of 44 weekly boards and has nine frontier models, all above $1 per answer. Of the 25 frontier models under $1, 19 are Chinese.
Getting started
Arena's advice is to treat any leaderboard as a snapshot and re-check model choices regularly. Cost-sensitive teams should look at the sub-$1 frontier, where Xiaomi and DeepSeek kept the best model at their price almost every month despite slipping in rank. Arena also projects, from Opus 4.7's price drop, that Opus 5.5 (Max)-level performance could cost under $1 per answer by April 2027; that is a projection, not a promise.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.