Moonshot AI: Kimi K3.1 Pro scores 55, cuts API prices 33%

Moonshot AI released Kimi K3.1 Pro, a reasoning model that currently holds the fourth position on the Artificial Analysis Intelligence Index. The system achieved a score of 55 points, an 11-point increase over its predecessor, Kimi K3. This score places it three points behind the top proprietary systems, specifically Anthropic's Claude Opus 5.5, which scored 58.

The company positioned K3.1 Pro as a cost-efficient alternative for enterprise users. API pricing dropped by 33 percent with the launch. Input tokens are now priced at $2 per million, while output tokens cost $10. Cache hits are available at $0.20 per million tokens. Although the model generates 14 percent more output tokens, the total cost to execute a standard Intelligence Index task is $1.98, nearly identical to the $2.00 cost of the previous generation.

Performance gains were most evident in agentic tasks. On the GDPval-AA benchmark, K3.1 Pro reached an Elo rating of 1,846. This figure is 312 points higher than K3 and trails only Claude Opus 5.5 and Muse Wonder Full. In terminal operations, the new model recorded a 65.7 percent score on Terminal-Bench 4.0, surpassing GPT-6 Astra's 58.2 percent. The previous K3 model managed only 13 percent on that specific test.

Moonshot also reported a decrease in hallucination rates. On its internal Kimi Hallu Bench Hard, claim-level hallucinations dropped from 23.4 percent in K3 to 8.3 percent. The company notes that GPT-6 Astra still leads this metric at 5.5 percent. Financial benchmarking showed gains of 7.2 points in modeling and 11.6 points in research. Moonshot states that K3.1 Pro outperforms all tested models on OfficeQA Pro v2.

The model maintains the same architecture as K3, featuring 2.8 trillion total parameters with 104 billion active. It supports a one-million-token context window and processes both text and images. A variant called K3.1 Pro Highspeed is available, which Moonshot claims runs two to three times faster. Both versions are accessible via Kimi.ai, Kimi Work, Kimi Code, and the API.

Moonshot plans to publish the model weights by November 4, 2026. If released on schedule, K3.1 Pro would become the highest-scoring open-weights model available. This would narrow the performance gap between open and proprietary systems from 12 points to three. The company adjusted the training data for the open release, removing specific cybersecurity content to balance offensive and defensive capabilities.

In cybersecurity tests, the uncleaned model solved 90.5 percent of tasks on the xdev-bench-hard benchmark. After cleaning, the rate dropped to 81.0 percent. DeepSeek-V4.1-Flash scored 85.7 percent on the same test. The new model also integrates with game engines such as Unity and Unreal Engine, and it can write scripts for Blender and CAD software without additional connectors.