Kimi K3 and Qwen3.8 Max Tie in Intelligence, Alibaba to Open Weights Next Week
rohanpaul_ai · x · 2026-08-07
According to Artificial Analysis, Kimi K3 still leads open weights at 57 on the Intelligence Index, but Alibaba's upcoming Qwen3.8 Max is just 1 point behind at 56, making them effectively tied on capability.
Parameters & Cost:
- Kimi K3: 2.8T total parameters, $0.86 per task.
- Qwen3.8 Max: 2.4T total parameters with 95B active, $1.14 per task.
In a shift of strategy, Alibaba announced plans to release the Qwen3.8 Max weights next week, moving away from keeping its Max class proprietary. Once released, it will be their largest open weights model to date. Although Qwen3.8 Max beats Kimi K3 on GDPval-AA (1739 vs 1685 Elo), it requires more agentic turns, driving up output token usage by 45% and resulting in a higher cost per task.
Related event: Kimi K3 and Qwen3.8 Max Evaluations Approach Top Closed-Source Models(2 posts)→
More from Models
- Users report Gemini's memories can't actually be deleted even after wiping chat history — ThrowRA_Purpose2025 · 2026-10-16
- OpenAI stealth-testing mysterious internal model 'Aeon' in Codex desktop app and Dots — cedric_chee · 2026-10-16
- Argon edges out astra as daily driver, but Claude still wins heavy coding work — iruletheworldmo · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- RoboVista benchmark: Gemini-4 Argon hits 81.9% on robotics understanding — berkeley_ai · 2026-10-16
- Dev reports decision model jev beats gpt-6-luna on quality in real training pipeline — genmon · 2026-10-16