Chinese AI model closes the gap with top global rivals

The race for artificial intelligence supremacy is tightening as a Chinese firm delivers a model that nearly matches the best proprietary systems in the world. Moonshot AI has released Kimi K3.1 Pro, a reasoning engine that has climbed to fourth place on the Artificial Analysis Intelligence Index. This marks a significant shift in the global landscape, where open-weight models are rapidly eroding the lead held by closed, commercial systems from the United States and elsewhere.

The new model scores 55 points on the index, an improvement of 11 points over its predecessor. It sits just three points behind the top proprietary systems, including Anthropic's Claude Opus 5.5 and Muse Wonder Full. If Moonshot follows through on its plan to publish the model weights by November 4, 2026, K3.1 Pro would become the highest-scoring open-weights model available. This move would reduce the performance gap between open and proprietary models from 12 points to just three, assuming no other changes occur in the meantime.

Performance gains are most visible in agentic tasks, where the model demonstrates superior ability to handle complex workflows. On the GDPval-AA benchmark, K3.1 Pro reached 1,846 Elo, surpassing its predecessor by 312 points and placing it just behind Claude Opus 5.5 and Muse Wonder Full. In terminal-based coding tests, the new version scored 65.7 percent on Terminal-Bench 4.0. This result beats GPT-6 Astra and closely trails Claude Opus 5.5, while the previous K3 model managed only 13 percent on the same test.

Moonshot also reports a substantial reduction in hallucinations, a persistent challenge in large language models. On its internal Kimi Hallu Bench Hard, claim-level hallucinations dropped from 23.4 percent in K3 to 8.3 percent in the new version. The company notes that GPT-6 Astra still leads this specific metric at 5.5 percent. Financial capabilities have also improved, with gains of 7.2 points in financial modeling and 11.6 points in financial research. Moonshot claims the model now tops all tested systems on the OfficeQA Pro v2 benchmark.

The release includes a Highspeed variant that runs two to three times faster, available through various Kimi platforms. The architecture remains unchanged from K3, featuring 2.8 trillion total parameters with 104 billion active and a one-million-token context window. API prices have fallen by 33 percent, with input costs at $2 per million tokens and output at $10. Despite generating 14 percent more output tokens, the total cost to run an Intelligence Index task is $1.98, nearly identical to the previous model's $2.00.

A key area of focus is cybersecurity, where the model's capabilities have grown due to scaled post-training. Moonshot states that cyber-related content was part of the general training corpus, boosting the model's ability to research vulnerabilities. For the open-weights release, the company removed this data to balance offensive and defensive potential. On an internal benchmark reproducing recent vulnerabilities in software like the Linux kernel, the uncleaned model solved 90.5 percent of tasks, compared to 81.0 percent after cleaning. DeepSeek-V4.1-Flash scored 85.7 percent, while the older Kimi K3 scored 33.3 percent.

Moonshot argues that open-weights models should not push the absolute frontier of offensive cyber capabilities but should gradually raise the bar to support defensive work such as auditing legacy systems. The company points to previous findings by Anthropic, which showed that even lower-scoring models can write exploits at levels close to top-tier proprietary systems. This release highlights the increasing competitiveness of non-US AI developers in a market long dominated by American firms.

The outlook suggests a more balanced global AI ecosystem where cost and performance are closely aligned. With the weights set to be released in early November, developers worldwide will have access to a top-tier model that challenges the status quo. The narrowing gap between open and proprietary systems indicates that the advantage of closed models may be diminishing, forcing major players to compete on innovation rather than just exclusivity.