🔍 Read the full analysis: A Closer Look At Kimi K3.1 Pro’s Near-Frontier AI Performance on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Artificial Analysis scored Moonshot AI’s Kimi K3.1 Pro at 55.1 on its Intelligence Index, placing it fourth among models in the October 24 comparison and three points behind the leading score. The model is available through Kimi services, but its weights are only promised by November 4 and Moonshot has not published a K3.1 licence.
Moonshot AI’s Kimi K3.1 Pro launched on October 23, and an evaluation by Artificial Analysis scored it 55.1 on the Intelligence Index, fourth in the comparison available the next day. The result puts the model three points behind the highest score in that table, while its weights remain promised for November 4 and the company has not published a licence for them.
Artificial Analysis lists K3.1 Pro as a mixture-of-experts model with 2.8 trillion total parameters and 104 billion active parameters. It accepts text and image input and has a listed one-million-token context window. Moonshot made it available through the Kimi apps, Kimi Work, Kimi Code and its API.
In the October 24 comparison, K3.1 Pro scored 55.1, behind Claude Opus 5.5 at 57.6, Claude Sonnet 5.5 at 56.0 and Muse Wonder Full at 55.8. Artificial Analysis reported a cost of $2.11 per Intelligence Index task for K3.1 Pro, compared with $5.98 for Opus 5.5 at its highest reasoning setting. The service’s listed API prices are $2 per million input tokens and $10 per million output tokens.
The benchmarks also show a substantial rise in agentic task performance compared with the earlier Kimi K3: K3.1 Pro scored 67% on Terminal-Bench 4.0, against 13% for K3 in the cited evaluation. On Humanity’s Last Exam, the score rose from 47% to 52%; Opus 5.5 scored 61%. These results describe performance on particular tests, not a guarantee of equivalent performance in everyday use.
Infographic 01 · Claimed vs. measured
Moonshot says “matches the strongest.” Independent tests say: close.
The KPIs at the top are independent measurements. Vendor numbers and independent numbers come from different harnesses, so they are shown side by side, never mixed.
Moonshot reports
vendor claims own harness, partly internal tests
| Benchmark | K3.1 Pro | Comparison |
|---|---|---|
| Terminal-Bench 4.0 | 65.7% | vs. Claude Opus 5.5: 66.4% |
| BenchCAD with tools | 92.7% | vs. Claude Opus 5.5: 96.2% |
| OfficeQA Pro v2 | 64.4 | vs. Claude Opus 5.5: 62.2 |
| Kimi Hallu Bench Hard internal | 8.3% | vs. Kimi K3: 23.4% (about −65%) |
Artificial Analysis measured
independent highest reasoning setting per model
| Measurement | K3.1 Pro | Comparison |
|---|---|---|
| Terminal-Bench 4.0 | 67% | vs. Claude Opus 5.5 (max): 60% |
| GDPval-AA | 1,846 Elo | vs. Claude Opus 5.5: 1,866 |
| Humanity’s Last Exam | 52% | vs. Claude Opus 5.5: 61% |
| AA-Omniscience hallucination rate | 44% | vs. Kimi K3: 53% · Gemini 4 Argon: 15% |
The 65% hallucination drop holds on Moonshot’s own stress test. On an independent knowledge test, the rate fell from 53% to 44%.Moonshot: the stress test rates “do not represent overall error rates in daily use.”
Sources: Moonshot AI, Kimi K3.1 Pro tech blog (23 Oct 2026) (left card, vendor charts); Artificial Analysis launch article (23 Oct 2026) and Artificial Analysis model page (data as of 24 Oct 2026) (KPIs and right card).
Powered by Thorsten Meyer AI
A Narrower Gap to Closed Models
The result matters because it places K3.1 Pro close to the leading proprietary models on one independent composite index, while its reported task cost is lower than Opus 5.5’s in that comparison. For organizations selecting AI services, that combination may make the model worth testing for coding, research and multi-step workflows, especially where API cost is a consideration.
The distinction between access and openness is important. Customers can use the model through Moonshot’s services now, but they cannot yet independently download and deploy its weights. Artificial Analysis describes K3.1 Pro as set to become the leading open-weights model if the weights are published. Until then, claims about self-hosting, inspection and deployment control remain conditional. The model’s 2.8-trillion total parameter count also means running it locally would require substantial infrastructure, according to the source report.
Popular right now
Bose QuietComfort Wireless Bluetooth Headphones – Active Over‑Ear Noise Cancelling with Mic, USB‑C Charging, Deep Bass, up to 24 h Playtime,
Hisense 55-Inch E6SR Series Hi-QLED 4K UHD Smart Roku TV (55E6SR, 2026 Model)
KitchenAid Artisan 5-Qt Tilt-Head Stand Mixer, KSM150PS, Almond Cream
As an affiliate, we earn on qualifying purchases.
From Kimi K3 to K3.1 Pro
K3.1 Pro is described as a more heavily post-trained version of Kimi K3. The prior K3 scored 43.6 on the Artificial Analysis Intelligence Index in the same table, while the earlier best open-weights entry cited in the report, Xiaomi’s MiMo-V2.6-Pro, scored 46.3. K3.1 Pro’s 55.1 therefore marks a sizable benchmark change, though the compared systems and test results do not establish that every user task improved by the same amount.
Moonshot has also promoted demonstrations involving game engines, Blender and computer-aided design, finance tasks, and reduced hallucinations. Those claims come largely from company demonstrations and internal tests. Artificial Analysis provides a separate evaluation: on its AA-Omniscience test, the reported hallucination rate for K3.1 Pro was 44%, down from 53% for K3. The source report says Moonshot’s internal test reported a larger reduction, from 23.4% for K3 to 8.3% for K3.1 Pro, but the company cautioned that the test was deliberately designed to provoke errors and does not represent overall error rates in daily use.
“The internal hallucination test is “deliberately crafted” to provoke errors, and its rates “do not represent overall error rates in daily use.””
— Moonshot AI
Weights, Licence and Real-World Results
Moonshot has said the weights will be released by November 4, but at the time of the October 24 report they were not available. The company had also not published a K3.1 licence, leaving unclear what users would be allowed to do with the weights, including whether commercial use or redistribution would be permitted.
Benchmark results do not settle how the model will perform across customer workloads. Moonshot’s demonstrations and internal scores are vendor claims, while Artificial Analysis tested a defined set of tasks and reported results from its own evaluations. The supplied material does not establish the methods behind every vendor test or show how performance varies across different prompts, tools and deployment settings. The reported cost per index task is also tied to the evaluation setup; it is not a universal estimate of operating cost.
The November 4 Release Test
The next milestone is Moonshot’s stated target to publish K3.1 Pro’s weights by November 4. The release and its accompanying licence will determine whether developers can run and adapt the model outside Moonshot’s hosted products, and under what conditions. If the date changes or the terms are restrictive, the practical meaning of the open-weights comparison will change.
Readers and prospective users can also watch for additional independent evaluations and for testing on their own workloads. Until then, the clearest confirmed picture is a strong score in Artificial Analysis’s October 24 comparison, lower reported task cost than Opus 5.5 in that test, and a still-pending release of weights and licence.
Key Questions
What is Kimi K3.1 Pro?
Kimi K3.1 Pro is a Moonshot AI model that the company describes as a more heavily post-trained version of Kimi K3. It launched through the Kimi apps, Kimi Work, Kimi Code and the Kimi API on October 23, 2026.
How did it rank in the reported benchmark?
Artificial Analysis scored it 55.1 on its Intelligence Index, placing it fourth in the October 24 comparison. The top score shown was 57.6 for Claude Opus 5.5.
Can users download and run the model now?
Not according to the supplied report. Moonshot had promised the weights by November 4, but they had not yet been published as of October 24. The company had not published a K3.1 licence either.
What does it cost to use through the API?
The listed price is $2 per million input tokens and $10 per million output tokens. Artificial Analysis estimated $2.11 per Intelligence Index task under its evaluation, a test-specific figure rather than a general cost for every use.
Are Moonshot’s performance and hallucination claims independently confirmed?
Not all of them. Artificial Analysis independently evaluated the model on its own tests, including reporting a 44% hallucination rate on AA-Omniscience, down from 53% for Kimi K3. Moonshot’s larger reduction figure came from an internal test, which the company said does not represent overall error rates in daily use.
Source: ThorstenMeyerAI.com
