Chinese AI startup Moonshot AI has launched Kimi K3.1 Pro, a post-trained reasoning model that scores 55 points on the Artificial Analysis Intelligence Index. The performance places it fourth overall, trailing proprietary frontier models such as Anthropic’s Claude Opus 5.5 and Claude Sonnet 5.5. Moonshot plans to release the model weights by November 4, 2026, which would significantly narrow the intelligence gap between open-weights and proprietary offerings.

The model demonstrates strong gains in agentic benchmarks, scoring 1,846 Elo on GDPval-AA and 65.7 percent on Terminal-Bench 4.0. K3.1 Pro features native integration with game engines like Unity and Unreal Engine, as well as direct script execution for CAD software. Moonshot simultaneously reduced API token costs by 33 percent while maintaining a one-million-token context window across text and vision capabilities.

Why it matters

  • Narrows the performance gap between proprietary frontier models and open-weights releases to its slimmest margin to date.

  • Provides engineers with native capabilities for 3D modeling and CAD script execution without requiring external connectors.

Source: the-decoder.com