Moonshot AI apologised for rate limits after its Kimi K3.1 Pro launch and gave subscribers a one-time reset restoring a quarter of their monthly quota

Overview Debates Related claims Topics Sources

441e85161News eventClaim

Moonshot AI released Kimi K3.1 Pro, a 3-trillion-parameter model trained to restrict its use in cyberattacks

Moonshot AI releases Kimi K3.1 Pro and says it cut the model's hacking skills before open release

Moonshot AI reported Kimi K3.1 Pro scored 65.7% on the Terminal-Bench 4.0 coding benchmark, against 66.4% for Anthropic's Claude Opus 5.5 and 58.2% for OpenAI's GPT-6 Astra

Moonshot AI promised to release the full Kimi K3.1 Pro model weights by November 4, 2026

Moonshot AI launched Kimi K3.1 Pro Highspeed alongside Kimi K3.1 Pro, running two to three times faster

Moonshot AI said it removed cyberattack-related data from Kimi K3.1 Pro's training before the model's open release

Moonshot AI released Kimi K3.1 Pro on October 23, 2026 through its Kimi apps and API

Kimi K3.1 Pro still scored far above Kimi K3's 33.3% on Moonshot AI's xdev-bench-hard exploit test despite the cyber-data removal

Kimi K3.1 Pro solved 81% of Moonshot AI's internal xdev-bench-hard exploit tasks, against 90.5% for a version trained without the cyber-data removal

Moonshot AI said Kimi K3.1 Pro's cyber capability remains behind frontier proprietary models that have been public for at least three months

At launch in October 2026, Kimi K3.1 Pro was priced at $2 per million input tokens and $10 per million output tokens on Moonshot AI's API

Kimi K3.1 Pro scored 55 on the Artificial Analysis Intelligence Index in October 2026

Moonshot AI called Kimi K3.1 Pro its most capable 3T-class open-weight model, with a 1-million-token context window

In Moonshot AI's hallucination stress test, Kimi K3.1 Pro made false claims in 8.3% of cases, down from 23.4% for Kimi K3

Open-weight models cannot be considered safe because their safeguards can be easily removed