Kimi K3 ranks second on AA-Briefcase, but each task costs about $10.57
airesearch12 · x · 2026-07-22
Kimi K3 is being highlighted as a strong agentic-work model: it ranks second on AA-Briefcase with an Elo of 1543, just behind Claude Fable 5 at 1574.
The catch is cost. The benchmark estimate puts Kimi K3 at about $10.57 per task, roughly 10× Kimi K2.6 and even above Opus 4.8 on this workload.
The post also notes:
- Kimi K2.6 jumped from 816 to 1543 for K3, a +727 Elo gain
- K3 reached 51% rubric pass rate, versus 56% for Fable 5
- Its analytical Elo was 1754, nearly matching Fable 5’s 1744
Related event: Kimi K3 Ranks Second on AA-Briefcase but with High Costs and Long Runtimes(6 posts)→
More from Models
- Fern architecture claims 10M-token context with predictions still improving — AccBalanced · 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
- One stray 'u' and Claude starts spelling like a Brit — cantrell · 2026-10-27
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27