Model · Kimi K3
Updated · By llmwise, AI-assisted.
Kimi K3 in English, without a Kimi account
Kimi K3 is Moonshot's long-context reasoning model. In llmwise you chat with it in English with no Kimi account: it's in the one-time free trial of 5 messages, and Kimi K3 gets up to 125 messages a month on Pro ($20). Every request goes through OpenRouter, asking only for hosts that don't store or train on prompts.
Kimi K3 in llmwise, plan by plan
Kimi K3 is one of the models in Free's trial, and every paid plan has it: Pro gets up to 125 Kimi K3 messages a month, as many as on Claude Sonnet 5.
| Plan | Price | Kimi K3 |
|---|---|---|
| Free | $0 | Not in the trial |
| Plus | $10 a month | Up to 62 a month |
| Pro | $20 a month | Up to 125 a month |
| Max | $50 a month | Up to 400 a month |
| Ultra | $100 a month | Up to 900 a month |
| Studio | $200 a month | Up to 2,000 a month |
Prices don't include tax, which is added where it applies and shown before you pay. A paid plan's month is one allowance shared by every model, so each monthly count is the most you get if all of it goes to that model. It renews each billing period; everyday models refill daily at 00:00 UTC. Long chats count more per reply. How pricing works.
Fair use is published too: $3.75 (Plus), $7.50 (Pro), $20 (Max), $42 (Ultra), $85 (Studio) of AI cost a month. At typical sizes every message on a plan fits under it; outsize messages and research runs use it up sooner. Every limit, explained.
Kimi K3 shares each plan's monthly allowance with the other models, so the more you use one, the fewer the others have left. The everyday models' 60 messages a day on Pro are separate.
On a paid plan, a $10 top-up adds up to 100 more Kimi K3 messages, valid 12 months.
Kimi K3's native context window is 1.05M tokens, but llmwise stops a chat at 200k tokens. Long chats cost more: past 64k tokens each reply counts as 2 Kimi messages, past 128k as 4.
Where your Kimi K3 messages go
llmwise doesn't call Moonshot directly. Every Kimi K3 message goes through OpenRouter, a service that routes requests to the companies hosting a model.
Each request asks OpenRouter only for hosts that, according to OpenRouter, don't store your prompts for their own use or train on them. It also skips hosts listed as running a 4-bit (fp4) build of the model.
The host that answers can be a hosting company rather than Moonshot, and handles your request under its own terms.
llmwise keeps your chats so you can come back to them. You can delete a chat, or every chat, at any time.
In our test runs of September 27, 28, 29 and October 2, 7, 8, 9, 15, 2026, OpenRouter sent Kimi K3's 50 prompts to 6 hosts (Together 16, Wafer 13, Phala 10, Fireworks 9, DigitalOcean 1, Morph 1). By default OpenRouter load-balances a model's requests across its hosts, weighing price and recent uptime (OpenRouter's provider routing docs), so the host can change from one message to the next.
Every company involved is listed in the Privacy Policy.
Kimi K3 or Claude Fable 5.1?
It's a common question, and we don't answer it with benchmark scores. Here's what differs in llmwise. On Pro you get about 4 Kimi K3 messages for every Fable one (125 against 31 a month). At their makers' API prices a typical Kimi K3 message costs about $0.0225, a Fable one about $0.0750. Kimi K3 gets a PDF's text; Fable reads the whole file, charts and layout included.
The honest way to choose is to ask both the same question: start on Kimi K3, then switch the picker to Fable in the same chat. Fable sees the whole conversation, Kimi's answer included.
| Model | On Pro | On Free | Context window | Images | PDFs | Reasoning | API price per 1M, in / out |
|---|---|---|---|---|---|---|---|
| Kimi K3Moonshot | 125/mo on Pro | No | 1.05M tokens | Yes | Text only | Yes | $3.00 / $15.00 |
| Claude Fable 5.1Anthropic | 31/mo on Pro | No | 1M tokens | Yes | Whole file | Yes | $10.00 / $50.00 |
Kimi K3's price is Moonshot's own, $3.00 in and $15.00 out per million tokens: the top of what its OpenRouter hosts charge.
More models served the same way
GLM 5.3 from Z.ai gets up to 250 messages a month on Pro, and DeepSeek V4 Pro up to 250. Both go through OpenRouter the same way as Kimi K3.
Also from primary sources: Moonshot's API pricing (Kimi K3 at $3 in and $15 out per million tokens, with a 1,048,576-token context). Kimi K3's model card (an open-weight, multimodal model). Moonshot's Kimi K3 repository (its weights released under the Kimi K3 License). OpenRouter's Kimi K3 page (the model served through OpenRouter). Read .
Kimi K3 outside China: questions
Can I use Kimi K3 in English?
Yes. Write in English, or any language you like, and Kimi K3 answers in it. llmwise's pages, settings and help are in English.
Do I need a Kimi account?
No. You sign in to llmwise with a link sent to your email: no password, no phone number and no ID. llmwise reaches Kimi K3 through OpenRouter, so there's no Moonshot account to make.
Is Kimi K3 free in llmwise?
It's in llmwise's one-time trial of 5 messages, with no card. After that it needs a paid plan: Pro ($20 a month) gets up to 125 Kimi K3 messages a month.
Who sees my Kimi K3 messages?
The host that runs your request does. That can be Moonshot or a hosting company, chosen by OpenRouter from the hosts that, according to OpenRouter, don't store prompts for their own use or train on them. Each host handles requests under its own terms.
Kimi K3 in English, with no Kimi account
It's in the 5-message free trial, and on Pro ($20) it's up to 125 a month.