SignalTech/AI
Kimi K3.1 Pro model weights are out
Kimi K3.1 Pro model weights released, signaling new open AI model availability.
Context
What happened: Kimi K3.1 Pro model weights were released as an open-weight, natively multimodal agentic model, the first in the K3.1 series, post-trained from Kimi K3 with joint reinforcement learning across hundreds of downstream domains.
Key numbers: 2.8T total parameters, 104B activated, 896 experts with 16 selected per token,
Why it matters: The release reports gains in coding, 3D creation, financial research, factual accuracy and agentic capabilities, with benchmark comparisons against Claude Fable 5.1, GPT-6 Astra, Claude Opus 5.5, DeepSeek-V4.1-Flash-1017 and GLM-5.3.
What changed
Kimi K3.1 Pro model weights were released as an open-weight, natively multimodal agentic model, the first in the K3.1 series, post-trained from Kimi K3 with joint reinforcement learning across hundreds of downstream domains.
Why it matters
The release reports gains in coding, 3D creation, financial research, factual accuracy and agentic capabilities, with benchmark comparisons against Claude Fable 5.1, GPT-6 Astra, Claude Opus 5.5, DeepSeek-V4.1-Flash-1017 and GLM-5.3.