Moonshot introduced Kimi K3.1 Pro as the first model in its K3.1 series and its most capable 3T-class open-weight model to date, with a 1-million-token context window. The company also launched Kimi K3.1 Pro Highspeed, which it says runs 2–3 times faster than the standard model.
Moonshot says both models are available on Kimi.ai, Kimi Work, Kimi Code and the Kimi API. It describes K3.1 Pro as a post-trained version of Kimi K3, developed with substantially more post-training compute, and reports a 65.7% score on Terminal-Bench 4.0, compared with 58.2% for GPT-6 Astra and 66.4% for Claude Opus 5.5.
The company says it removed cyber-related data from the mid-training corpus and filtered cyber-related tasks from the RL environment mix for the open-weight release, while retaining general-purpose coding tasks. It reports an 81.0% pass@3 result on its internal xdev-bench-hard benchmark for K3.1 Pro, versus 90.5% for a checkpoint without those risk-reduction measures; the page excerpt ends before giving further details about the final result.
The long context window and reported coding results could make the model relevant to builders developing coding agents and other tool-using systems; however, benchmark results are company-reported and may not predict performance in real-world use.