Topic

Kimi K3

3 stories

Ubicloud Achieves 100% Tool-Call Accuracy Using Constrained Decoding

Ubicloud improved tool-call schema accuracy for Kimi and DeepSeek models from 92% to 100% by enabling constrained decoding in a vLLM fork.

AIOct 27, 12:22 AM UTC1 source

Cerebras Deploys Trillion-Parameter Models on Wafer-Scale Chips

Cerebras has deployed large AI models using its WSE-3T wafer-scale chips, achieving high inference speeds through a multi-wafer pipeline architecture.

AIOct 25, 10:13 PM UTC1 source

Moonshot AI releases Kimi K3.1 Pro with 2.8 trillion parameters

Chinese AI company Moonshot AI has released the Kimi K3.1 Pro model, which ranks fourth on the Artificial Analysis Intelligence Index.

AIOct 25, 10:15 AM UTC2 sources