Moonshot's New AI Model Narrows the Gap With Industry Leaders

The competitive landscape for large language models has shifted dramatically in the last few days. A new release from a Chinese tech firm has moved into the upper echelon of global benchmarks, challenging the dominance of established American rivals.

Moonshot AI has officially launched Kimi K3.1 Pro, a post-trained variant of its previous iteration. According to the Artificial Analysis Intelligence Index, the new model scores 55 points. This represents an 11-point increase over its predecessor, Kimi K3, and places it in fourth position globally.

It is worth noting that Kimi K3.1 Pro is currently the fourth-highest scoring model on the index. It trails only three proprietary systems: Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full. The gap between the new release and the top performer is just three points.

The timeline for this development is quick. The model was released recently with a focus on enhanced reasoning capabilities. If the company follows through on its plan to release the model weights by November 4, 2026, Kimi K3.1 Pro will become the highest-scoring open-weights model available. This would reduce the performance gap between open and proprietary models from 12 points to just three, assuming no other major releases occur in the interim.

Performance improvements are most visible in agentic tasks. On the GDPval-AA benchmark, the new model achieves an Elo rating of 1,846, a jump of 312 points from the previous version. It also posts a 65.7 percent score on Terminal-Bench 4.0, a task where the older model managed only 13 percent. These figures place it close to top competitors like Claude Opus 5.5 and GPT-6 Astra.

Beyond coding, the model shows strong capabilities in 3D modeling and finance. It can interact directly with game engines such as Unity and Unreal Engine. On the BenchCAD benchmark, it scores 92.7 percent with tools, nearly matching the industry leader. Moonshot also reports significant gains in financial modeling and research tasks, with the model topping the OfficeQA Pro v2 test for general office work.

Pricing has been adjusted to reflect the increased efficiency. While the model retains 2.8 trillion total parameters, API costs have dropped by 33 percent. Input tokens now cost $2 per million, down from $3, while output tokens are priced at $10, down from $15. The company states that this keeps the cost per intelligence task roughly the same as the previous version, despite the model generating more output tokens.

A notable feature of the release is the handling of cybersecurity capabilities. Moonshot noted that vulnerability research skills increased naturally during training. For the open-weights version, the company removed specific cyber-related data from the training set. This reduces the model's ability to solve complex security tasks from 90.5 percent to 81.0 percent on an internal benchmark, a deliberate choice to balance offensive and defensive utility.

The next steps for Moonshot are clear. The company aims to release the full model weights by early November 2026. In the meantime, both the standard Kimi K3.1 Pro and a faster variant, K3.1 Pro Highspeed, are available for immediate use. This faster version operates two to three times quicker than the standard release, offering users a choice between maximum depth and rapid execution.

This release underscores the rapid pace of innovation in the AI sector. By combining significant performance gains with lower costs, Moonshot is positioning its new model as a serious alternative to leading proprietary systems. The coming months will determine if the open-weights community can sustain this momentum as more models are released.