Kimi K3.1 Pro Hits 55 Points, Narrowing Gap With Top AI Rivals

Moonshot AI’s newly released Kimi K3.1 Pro has surged to fourth place on the Artificial Analysis Intelligence Index, scoring 55 points and closing the distance to the industry leaders. This 11-point increase over its predecessor, Kimi K3, places the model just three points behind the top proprietary systems. The primary driver of this improvement is a substantial expansion in post-training compute, which has refined the model’s reasoning capabilities without altering its fundamental architecture.

For Midwest readers following the AI landscape, the most notable feature of this release is its aggressive pricing strategy. While the model retains the same 2.8 trillion total parameters and 104 billion active parameters as the previous version, the cost to use it has dropped by 33 percent. Input tokens now cost $2 per million, down from $3, while output tokens are priced at $10 per million, a reduction from $15. Cache hits have also seen a price cut to $0.20 per million tokens. Despite the lower rates, the cost per task on the Intelligence Index remains nearly identical to the previous model at $1.98, primarily because the new version generates 14 percent more output tokens during standard evaluations.

The technical upgrades are most evident in agentic tasks and coding. On the GDPval-AA benchmark, K3.1 Pro reached an Elo rating of 1,846, a jump of 312 points from its predecessor. It also demonstrated significant progress on Terminal-Bench 4.0, achieving a success rate of 65.7 percent. This performance now rivals GPT-6 Astra and sits only slightly below Claude Opus 5.5. Additionally, the model’s hallucination rates have decreased sharply; on Moonshot’s internal stress test, claim-level errors fell from 23.4 percent to 8.3 percent, though GPT-6 Astra still holds a slight edge at 5.5 percent.

In specialized fields, the model shows particular strength in 3D modeling and finance. It can interact directly with game engines like Unity and Unreal Engine, and it scored 92.7 percent on the BenchCAD benchmark when using external tools. Moonshot also introduced a feature called 3D Make on its platform, allowing users to preview and export models. In financial applications, the model gained 7.2 points on Financial Modeling and 11.6 points on Financial Research, topping all tested competitors on the OfficeQA Pro v2 benchmark for general office tasks.

Moonshot AI is preparing to release the full model weights by November 4, 2026. If this happens as scheduled, K3.1 Pro would become the highest-scoring open-weights model available. This move would shrink the performance gap between open and proprietary models from 12 points to just three, assuming no other major releases occur in the interim. The company also noted that it has cleaned cyber-related data from the training process to balance offensive and defensive capabilities for the open-weights version.

Currently, both the standard K3.1 Pro and a faster variant called K3.1 Pro Highspeed are available through Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. The Highspeed version operates two to three times faster than the standard model. For developers and businesses in the region, this release represents a significant step toward more affordable and capable AI infrastructure, particularly for those who rely on open-weights models for customization and cost management.