Moonshot AI Releases Kimi K3.1 Pro: A 2.8T Total/104B Active Params (MoE) Model with 1M Context

Moonshot AI has unveiled Kimi K3.1 Pro, a mixture-of-experts model that totals 2.8 trillion parameters but activates only 104 billion per inference pass. The architecture allows the model to scale capacity without proportionally increasing compute cost, a strategy increasingly common among frontier labs.

The model supports a 1-million-token context window, positioning it for long-document analysis, extended agentic workflows, and multi-step reasoning tasks that require sustained attention over large inputs. It reportedly scores 65.7% on Terminal-Bench 4.0, a benchmark focused on terminal and coding agent performance.

Pricing is set at $2 per million input tokens and $10 per million output tokens, which is competitive for a model of this scale. Moonshot AI has indicated that open weights will be made available by November 4, 2026, which could enable independent evaluation and community fine-tuning.

Moonshot AI is a Chinese AI startup best known for the Kimi series of large language models. The K3.1 Pro release extends that lineage into the ultra-large MoE territory, where total parameter counts far exceed active ones. Whether the model's benchmarks translate into real-world gains for developers remains to be seen once open weights are released.

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.