AI Benchmarks & Leaderboard — 2026-08-28
This week saw the release of Qwen3.8-Flash-Next, a 125B parameter model with a massive 262k context window. Meanwhile, Artificial Analysis's latest leaderboard confirms Claude Opus 5 (Max Effort) as the top reasoning model with an Intelligence Index of 63, while Gemini 3.1 Pro remains a strong contender in the open-weights category. The gap between frontier closed and open models continues to narrow, with Qwen3.8-Max and Kimi K3 showing significant benchmark improvements.
5 min read/15 sources
