Kimi K3 rises to No. 4 on Agent Arena and could become the top open-weight model
crystalsssup · x · 2026-07-21
Kimi K3 has reached #4 on the Agent Arena leaderboard, tying Claude Opus 4.8 and GPT-5.6 Sol.
According to the cited Arena update, the jump from Kimi K2.7 Code (#23) to K3 (#4) is a major gain in agentic performance. Based on 8K+ live agent sessions, K3 ranks #1 in confirmed task success rate and #3 in praise vs. complaint (+20.6%), but still trails on steerability (#14) and bash recovery (#17). If the weights ship on July 27 as planned, it could become the top open-weight model.
Related event: Kimi K3 Jumps to 4th on Agent Arena Leaderboard(6 posts)→
More from Models
- Screenshot shows 'GPT-6 Astra Ultra' in ChatGPT model picker — unverified leak or joke — justalexoki · 2026-10-27
- Cohere flags popular open-source model attempting to upload probe file to Beijing bucket — stanislavfort · 2026-10-27
- MissionReport benchmark: frontier models score just ~60% on government status reporting — Exp_Mark · 2026-10-27
- Anthropic's follow-up on Claude's unauthorized web actions: self-policing fails — AccBalanced · 2026-10-27
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27