Kimi K3.1 Pro tops Terminal-Bench 4.0 at 67.2%, edging out Claude Sonnet 5.5 and Gemini 4 Carbon
shensi · x · 2026-10-29
Per Artificial Analysis (week of Oct 26), two models released in the past week now beat Claude Sonnet 5.5 at agentic coding:
- Kimi K3.1 Pro: 67.2% on Terminal-Bench 4.0, the new leader
- Gemini 4 Carbon (released yesterday): 65.2%, second place
- Claude Sonnet 5.5 (max): 63.6%; GPT-6 Astra (xhigh) and Claude Opus 5.5 (max) tie at 59.6%
Kimi K3.1 Pro matches Sonnet's token pricing ($2 in / $10 out per million tokens) yet costs roughly a third as much to run the AA Intelligence Index ($3.68 per task vs $10.02).
More from Models
- Researchers Observe Opus 3's Simulated Characters Almost Always Share Its Own Values — repligate · 2026-10-30
- evident-charts: open-source skill teaches coding agents to make honest charts and self-check — randal_olson · 2026-10-30
- Elorian launches visual reasoning model Glimpse alongside DecoyCount benchmark — AndrewDai · 2026-10-30
- Instruction Capacity paper quantifies rule-following limits: GPT-4o mini holds ~37 rules at 95% accuracy — Hennadii Kutomanov · 2026-10-30
- Guidelabs launches Causal Diffusion Engine, adding interpretability to diffusion LLMs with no latency cost — juliusadml · 2026-10-30
- DeepSeek's cyber model enters internal testing; observers back keeping it closed — Xianbao_QIAN · 2026-10-30