🔍 Read the full analysis: How Near The Frontier Is Kimi K3.1 Pro? Three Points Away on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Artificial Analysis gives Moonshot AI’s Kimi K3.1 Pro an Intelligence Index score of 55.1, placing it fourth and 2.5 points behind the leading score in its October 24 table. The model is available through Kimi services, but its weights are only promised by November 4 and Moonshot has not published a licence for them.
Moonshot AI launched Kimi K3.1 Pro on October 23, and an evaluation by Artificial Analysis ranked it fourth on its Intelligence Index, with a score of 55.1. That puts it 2.5 points behind the table’s leading score, Claude Opus 5.5 at 57.6; Kimi’s model is available through the company’s services, but its weights are not yet released and its licence has not been published.
Artificial Analysis lists K3.1 Pro as a mixture-of-experts model with 2.8 trillion total parameters and 104 billion active parameters. It supports a 1-million-token context window and text and image input. Moonshot made it available through the Kimi apps, Kimi Work, Kimi Code and its API. The firm describes it as a more heavily post-trained version of Kimi K3.
In Artificial Analysis’s October 24 table, K3.1 Pro’s 55.1 Intelligence Index score sits below Opus 5.5 at 57.6, Claude Sonnet 5.5 at 56.0 and Muse Wonder Full at 55.8. The scores reflect each model at its highest listed reasoning setting. Artificial Analysis also reports a cost of $2.11 per Intelligence Index task for Kimi, compared with $5.98 for Opus 5.5. These are evaluation-specific cost estimates, not a guarantee of what every user or workload will pay.
The same evaluation records a sharp increase on Terminal-Bench 4.0: K3.1 Pro scored 67%, compared with 13% for Kimi K3. On GDPval-AA, a simulated knowledge-work evaluation, it scored 1,846 Elo, behind Opus 5.5 at 1,866 and Muse Wonder Full at 1,857. Its score on Humanity’s Last Exam was 52%, up from 47% for K3; Opus 5.5 scored 61%.
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 for Open Weights
The result matters because it places a model whose weights Moonshot says it plans to release close to leading proprietary systems on one broad benchmark. If the weights are published on schedule and under terms that permit the uses developers want, teams may gain another option for running or adapting a capable model rather than relying solely on a hosted service. That could matter for companies with specific control, deployment or data-handling needs.
The distinction between a model score and an available open-weight system is central. K3.1 Pro can currently be accessed through Moonshot’s products and API, but organizations cannot yet download its weights. Artificial Analysis describes it as set to become the leading open-weights model once those weights are published; that is a conditional assessment, not a description of its current availability. Its 55.1 score is also an index result, not proof that it will outperform other models on every coding, research or business task.
Price comparisons also depend on the workload and evaluation method. Artificial Analysis estimates about $2.11 per index task for Kimi and about $5.98 for Opus 5.5 in its table. That points to a lower estimated cost for that test, but does not establish a universal three-to-one price advantage or account for every deployment expense, including infrastructure and operations.
Popular right now

VIZIO 65-Inch V-Series 4K HDR Smart TV

Dyson V15 Detect Origin Cordless Vacuum

De'Longhi La Specialista Arte Evo Semi-Automatic Espresso Machine with Grinder and Cold Brew
As an affiliate, we earn on qualifying purchases.
What the Benchmarks Measure
Moonshot’s launch presentation includes claims about building projects across Unreal Engine, Unity and Godot, generating parametric 3D models with Blender and CAD tools, and improving financial work. These are vendor claims and demonstrations, not results independently established by the Artificial Analysis index. Moonshot reports a 92.7% result on BenchCAD with tools, below its cited scores for Claude Opus 5.5 at 96.2% and GPT-6 Astra at 95.9%.
Moonshot also says its internal Kimi Hallu Bench Hard test showed claim-level hallucinations falling from 23.4% for Kimi K3 to 8.3% for K3.1 Pro, a reduction of about 65% on that test. The company cautions that the test was deliberately designed to provoke errors and that its rates do not represent everyday use. Artificial Analysis’s separate AA-Omniscience evaluation recorded a smaller decline, from 53% for K3 to 44% for K3.1 Pro. The different results come from different tests and should not be treated as interchangeable measures.
The launch follows a notable gap in the Artificial Analysis rankings. The source reports that Xiaomi’s MiMo-V2.6-Pro scored 46.3, while Opus 5.5 scored 57.6. K3.1 Pro’s 55.1 score is therefore much nearer the closed-model leaders on this index than the open-weight models cited in that comparison. Rankings vary by benchmark, and the index alone does not settle how models compare across all real-world tasks.
“Kimi K3.1 Pro is its most capable “3T-class open-weight model.””
— Moonshot AI
Weights and Licence Still Pending
The model’s most consequential open question is whether Moonshot will publish the weights by its stated November 4 target. The source material says the company has promised a release by that date, but reports that no K3.1 licence has been published. Without the licence, users cannot judge which commercial, research or redistribution uses would be permitted. Until the weights are available, the model should not be described as downloadable or independently self-hostable.
Practical deployment is also uncertain. The listed 2.8 trillion total parameters make local operation a substantial infrastructure undertaking, and the source characterizes running it independently as a data-centre-scale project. Moonshot’s demos and internal results have not been independently verified in the cited material. It is also unclear how the model will perform on each organization’s own tasks, costs and reliability requirements. Benchmark results offer evidence about defined tests, not a complete forecast of production performance.
Watch for the November Release
The next milestone is Moonshot’s promised November 4 weight release. Its timing, the publication of a licence and the licence’s exact terms will determine whether developers can download, modify and deploy K3.1 Pro, and under what conditions. If the release occurs, independent tests of the weights could help confirm whether the hosted model’s benchmark performance carries over to self-managed use.
Until then, readers can distinguish what is already available from what remains prospective: K3.1 Pro is live through Moonshot’s apps and API, while its open-weight status depends on a future release. Further comparisons should also be read against the specific benchmarks and settings used, rather than as a single verdict on performance across all tasks.
Key Questions
How close is Kimi K3.1 Pro to the top-ranked model?
Artificial Analysis scored it 55.1 in its October 24 table. The leading score was Claude Opus 5.5 at 57.6, a difference of 2.5 points.
Can users download Kimi K3.1 Pro now?
No. It is available through Moonshot’s Kimi apps, Kimi Work, Kimi Code and API, but the weights have not been released. Moonshot has promised them by November 4.
Has Moonshot published the model’s licence?
The source material reports that no K3.1 licence has been published. The permitted uses will not be clear until Moonshot releases the licence terms.
Does the model’s 65% hallucination reduction apply to everyday use?
That figure comes from Moonshot’s internal Kimi Hallu Bench Hard test. The company says the test was designed to provoke errors and does not represent overall error rates in daily use. Artificial Analysis reported a smaller decline on its separate evaluation, from 53% to 44%.
Source: ThorstenMeyerAI.com
