Kimi K3 pricing and specs
Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work
Kimi K3 price per 1M tokens
| Type | ₽ / 1M | $ / 1M |
|---|---|---|
| Input tokens | 254 ₽ | $3 |
| Output tokens | 1 268 ₽ | $15 |
| Cache read | 25,35 ₽ | $0.3 |
Prices in rubles as of 18.09.2026 at the CBR rate: 1 $ = 84.51 ₽.
Kimi K3 specs and limits
- Context window
- 1.0M
- Max output
- 131K
- Input
- text, image, video
- Output
- text
- Family
- kimi-k3
- Released
- 2026-07-16
- Updated
- 2026-07-16
What Kimi K3 can do
- ✓Reasoning
- ✓Tool calling
- ✓Structured output
- ✓Attachments / files
- ✓Vision (images)
- —Temperature control
- ✓Open weights
Kimi K3 benchmarks
| Benchmark | Score | Metric |
|---|---|---|
| DeepSWE | 67.5 | resolve rate |
| Terminal-Bench | 88.3 | accuracy |
| FrontierSWE | 81.2 | dominance score |
| Program Bench | 77.8 | score |
| SWE Marathon | 42 | resolve rate |
| GDPval-AA | 1668 | Elo |
| AA-Briefcase | 1548 | Elo |
| AutomationBench | 30.8 | success rate |
| JobBench | 52.9 | score |
| SpreadsheetBench | 34.8 | score |
| BrowseComp | 91.2 | accuracy |
| CharXiv Reasoning | 91.3 | accuracy |
| ZeroBench | 41 | pass@5 |
Frequently asked questions
How much does the Kimi K3 API cost?
Input costs $3 per 1M tokens (254 ₽) and output costs $15 per 1M (1,268 ₽). Moonshot AI bills the prompt and the completion separately, so the price of a call depends on how long both are. Ruble figures use the Bank of Russia rate of 84.51 ₽ per dollar for September 2026.
How much do 1,000 Kimi K3 tokens cost?
List prices are quoted per 1M tokens, so 1,000 tokens cost a thousandth of that: divide $3 for input and $15 for output by 1,000. The same rates in rubles are 254 ₽ and 1,268 ₽ per 1M tokens. Both prompt tokens and completion tokens are billed.
How many tokens does Kimi K3 hold?
The Kimi K3 context window is 1.0M tokens (1,048,576). Everything shares that budget: the system prompt, the dialogue history, attached files and the answer the model writes. The figure comes from the Moonshot AI model card as of September 2026.
What are the Kimi K3 limits?
The context window is 1.0M tokens (1,048,576) and a single response is capped at 131K tokens, so longer output has to be generated in parts. Rate limits are set by Moonshot AI per account and depend on your plan rather than on the model, so they are not listed here.
How is the Kimi K3 price in rubles calculated?
Providers quote their tariffs in US dollars, so the ruble amounts here are a conversion at the official Bank of Russia rate of 84.51 ₽ per dollar. For Kimi K3 that gives 254 ₽ per 1M input tokens and 1,268 ₽ per 1M output tokens. When the rate moves, the ruble price moves with it.
How is Kimi K3 different from other Kimi models?
Kimi K3 belongs to the Kimi K3 family at Moonshot AI. Models in the Kimi line differ in context size, price per 1M tokens and supported capabilities, so compare them by the numbers on their cards rather than by the name. Every Kimi model is listed on the provider page.
Can Kimi K3 reason?
Yes, Kimi K3 declares a reasoning mode: it works through intermediate steps before it answers. Those steps add output tokens, so a call with a long reasoning chain costs more and takes longer than a plain completion. Capabilities declared for the model: reasoning, tool calling, image input, structured output.
What capabilities does Kimi K3 support?
The declared capabilities of Kimi K3 are: reasoning, tool calling, image input, structured output. It accepts text, images, video as input, and every attachment consumes tokens from the shared context window of 1.0M tokens. The list comes from the Moonshot AI model card and is refreshed with the catalog as of September 2026.
Does Kimi K3 have open weights?
Yes, Kimi K3 ships with open weights, so it can be downloaded and served on your own hardware instead of being used only through the Moonshot AI API. What you may do with it is set by the license shown on the card. Catalog prices cover hosted API access.