Physix Frontier badgePhysix Frontier
PublishChinese | EN

Physix Frontier · News Briefing Card (IT Home · Oct 30, 2026)

Mistral opens 1T-parameter Large 4 weights, cuts coding token use

KEY FACTS

  • Mistral AI released the weights of Mistral Large 4 on October 29.
  • The model has 1 trillion total parameters and 49 billion activated parameters, and is natively multimodal.
  • The model was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in the company's own European data centers.
  • It scores 48% on the coding agent index, on par with GLM-5.3, while consuming 47% fewer tokens per task.
  • The model is released under a modified MIT license; MaaS resale requires a separate commercial license.

KEY DATA

1 trillionTotal parameters
49 billionActivated parameters
3,800GPU count
48%Coding agent index score

PHYSIX OBSERVATION

Trillion-parameter open models are shifting the competitive focus from benchmark scores to inference cost. ML4 matches GLM-5.3 with nearly half the token consumption, showing that the cost-performance battle in coding agents has already begun. The modified MIT license puts a gate on cloud resale, a compromise between open-source influence and commercial monetization: small and mid-sized teams can use it for free, but the bar for doing big business remains.

Source: IT Home report