Prototype · closed beta
unOpen.AI
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Model families
Kimi.
Open weights at frontier scale.
Moonshot AI publishes the weights of some very large Kimi models and keeps its newest flagships API-only at first. Running the open releases yourself is a serious hardware decision.
All Model Families →
Current release
Kimi today.
Moonshot AI ships these models as a range rather than one lead model, so this page is about the family: what it is good at, where it runs and how we evaluate it.
Last reviewed on 30 October 2026. We check the public model catalogues every week; when a new release appears, we review it against the provider's announcement and update this page.
LATEST MODELS
The newest Kimi models.
What Moonshot AI has shipped most recently. Each has its own page with specifications, independent ratings at every reasoning setting and listed prices.
23 Oct 2026
Kimi K3.1 Pro
- Intelligence
- 46.0
- Context
- 1.05M
- In / out per M
- $2.00 / $10.00
16 Jul 2026 · open weights
Kimi K3
- Intelligence
- —
- Context
- 1.05M
- In / out per M
- $0.192 / $15.00
12 Jun 2026 · open weights
Kimi K2.7 Code
- Intelligence
- —
- Context
- 262k
- In / out per M
- $0.600 / $3.50
20 Apr 2026 · open weights
Kimi K2.6
- Intelligence
- —
- Context
- 262k
- In / out per M
- $0.650 / $3.41
EVERY MODEL
All 8 Kimi models, compared.
Sort by any column. Select a model for its full page.
| Kimi K3.1 Pro | 23 Oct 2026 | 46.0 | 30 | 1.05M | $2.00 | $10.00 |
|---|---|---|---|---|---|---|
| Kimi K3 open | 16 Jul 2026 | — | — | 1.05M | $0.192 | $15.00 |
| Kimi K2.7 Code open | 12 Jun 2026 | — | 48 | 262k | $0.600 | $3.50 |
| Kimi K2.6 open | 20 Apr 2026 | — | — | 262k | $0.650 | $3.41 |
| Kimi K2.5 open | 27 Jan 2026 | — | — | 262k | $0.500 | $2.50 |
| Kimi K2 Thinking open | 6 Nov 2025 | — | — | 262k | $0.600 | $2.50 |
| Kimi K2 0905 open | 5 Sep 2025 | — | — | 262k | $0.600 | $2.50 |
| Kimi K2 0711 open | 11 Jul 2025 | — | — | 131k | $0.570 | $2.30 |
Intelligence and coding: Artificial Analysis indices at the model's best reasoning setting, and its measured output speed. Context and prices: the OpenRouter catalogue list price in USD per million tokens. Released: the provider's date where Artificial Analysis records one, otherwise the date the catalogue first listed it. Refreshed weekly; not measured by QQuantum.ai. Select a heading to sort.
Specification
- Provider
- Moonshot AI, Beijing, China
- Weights
- Published on Hugging Face (moonshotai) for releases such as Kimi K3 and K2.6; newer flagships such as K3.1 Pro are API-only at the time of writing
- Licence
- Differs by release: Kimi K3 carries Moonshot's own Kimi K3 licence; K2.6 a modified MIT licence. Read the model card of the release you deploy
- Scale
- Kimi K3's model card lists roughly 2.8 trillion parameters: self-hosting it needs a multi-node GPU deployment
- Access paths
- Moonshot's API, third-party model catalogues, or an engineered deployment of the published weights
Provider-published specifications as of 30 October 2026, not figures we measured. Model families move quickly — check the provider's own documentation before committing to a number.
INDEPENDENT RATINGS
How Kimi compares.
Third-party scores and listed prices, refreshed every week. Useful for building a shortlist; your own evaluation decides.
Composite index across their reasoning, knowledge, maths and coding evaluations · Higher is better · Linear scale from 0 to 66.0
- Gemini 4 Carbon
- 59.4
- Claude Opus 5.5
- 58.4
- GPT-6 Astra
- 56.9
- Muse Wonder Full
- 49.7
- Kimi K3.1 Pro
- 46.0
- Kimi K3
- 17.4
This family's highest-rated models, highlighted, against the best-rated model of each other family we cover. One composite score cannot show cost, latency, data residency or how a model handles your task.
Source: Artificial Analysis. Checked 30 October 2026. Measured independently by Artificial Analysis, not by QQuantum.ai or the providers.
USD per million output tokens · The newest releases in the family · Linear scale from 0 to $17.00
- Kimi K3.1 Pro
- $10.00
- Kimi K3
- $15.00
- Kimi K2.7 Code
- $3.50
- Kimi K2.6
- $3.41
- Kimi K2.5
- $2.50
Input is priced separately: Kimi K3.1 Pro $2.00, Kimi K3 $0.19, Kimi K2.7 Code $0.60, Kimi K2.6 $0.65, Kimi K2.5 $0.50 per million tokens. Catalogue list prices; the provider's own rates, caching and batch discounts can differ.
Source: OpenRouter model catalogue. Checked 30 October 2026. Listed by OpenRouter, not measured by QQuantum.ai.
Where we would choose something else
Open weights at this size do not make self-hosting cheap. Unless volume justifies a multi-node deployment, a hosted endpoint or a smaller open model is usually the better economics, and a replay evaluation shows how much quality the smaller model gives up on your task.
Where policy rules out models from Chinese providers, Mistral and Meta's models are the adjacent open options. Where it does not, separate the hosted API, which sends data to the provider, from published weights on your own hardware, which do not.
Questions
Questions about Kimi by Moonshot AI.
Can we run Kimi on our own hardware?
The releases with published weights can be self-hosted under their licence terms, but the largest need a multi-node GPU deployment. Size weights, context cache and concurrency together, and confirm runtime support for the exact release before buying hardware.
Is the newest Kimi model open?
Not always at launch: Moonshot has kept recent flagships API-only at first. The table above marks which models have published weights.
What does the provenance review involve?
For self-hosting: pin the artefact, keep its licence, inspect the serving stack for outbound connections, and test behaviour on the topics your users raise. For the hosted API: review processing locations, retention and terms. Our model supply-chain approach covers both.
