Moonshot AI cuts prices 33% with Kimi K3.1 Pro

API costs for one of the world's leading reasoning models dropped by 33 percent this week as Moonshot AI launched Kimi K3.1 Pro. The price reduction coincides with a significant performance boost that places the model within striking distance of the market leaders in the proprietary sector.

The new release scores 55 points on the Artificial Analysis Intelligence Index. This 11 point increase over its predecessor, Kimi K3, propels the model into fourth place globally. It trails only Claude Opus 5.5 with 58 points, Claude Sonnet 5.5 with 56 points, and Muse Wonder Full with 56 points. The primary driver of this jump is additional post-training compute applied to the existing architecture.

Market analysts are watching the open-weights segment closely. If Moonshot follows through on its plan to release model weights by November 4, 2026, K3.1 Pro would become the highest-scoring open-weights model available. Such a move would narrow the performance gap between open and proprietary systems from 12 points to just three, assuming no other major releases occur in the interim. This shift could pressure incumbents to adjust their own pricing and performance strategies.

Performance gains are most visible in agentic tasks and coding. On the GDPval-AA benchmark, the model achieves 1,846 Elo, a 312 point leap from the previous version. It also records 65.7 percent on Terminal-Bench 4.0, a test where its predecessor managed only 13 percent. These results place it just behind Claude Opus 5.5 and ahead of GPT-6 Astra in several specific metrics. The company also reports a sharp decline in hallucination rates, dropping from 23.4 percent to 8.3 percent on its internal stress test.

Technical specifications remain consistent with the previous iteration. The model retains 2.8 trillion total parameters with 104 billion active during inference. It supports a one million token context window and processes both text and images. According to Artificial Analysis, it sits at the top of the Pareto frontier, meaning no model with a similar parameter count currently scores higher on the intelligence index.

Pricing changes reflect the efficiency gains. Input costs fall from $3 to $2 per million tokens, while output costs drop from $15 to $10. Cache hit prices decrease from $0.30 to $0.20. Despite lower per-token rates, the total cost per task remains nearly identical at $1.98, up from $2.00 for the previous model. This slight increase is due to the new model generating 14 percent more output tokens per task.

Moonshot has also made deliberate adjustments to the model's cybersecurity capabilities. To prepare for the open-weights release, the company removed specific cyber-related data from its training corpus. This cleaning process reduced the model's ability to solve recently disclosed vulnerabilities from 90.5 percent to 81.0 percent on internal benchmarks. The company states this balance aims to preserve offensive potential while ensuring open models do not push the absolute frontier of cyber capabilities too quickly.

A faster variant named K3.1 Pro Highspeed is also available, running two to three times quicker than the standard release. Both models are accessible via Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. The company has also introduced a 3D Make feature on its website that allows users to preview and export 3D models, capitalizing on the model's improved performance in 3D modeling and game engine integration.

Open questions remain regarding how competitors will respond to the shrinking gap between open and proprietary performance. The November deadline for the weight release is critical for validating the current benchmark standings. For now, the market sees a stronger competitor in the open-weights space that offers substantial cost savings without a significant drop in overall intelligence metrics.