Kimi API pricing for every model, newest first
By TechCompare · Updated · How we check
An 80,000 token prompt with a 4,000 token answer and half the input cached costs $0.128 on Kimi K3.1 Pro, the newest Kimi. Kimi K2.6, the lowest cost model Moonshot sells itself, runs it for $0.0604.
Every Kimi model and its API price
Newest first. Prices are per 1M tokens. One request is 80K tokens in and 4K out, with 50% of the input cached.
| Model | Input | Output | Cached | Context | One request | Listed |
|---|---|---|---|---|---|---|
| Kimi K3.1 ProMoonshot AI | $2 | $10 | $0.20 | 1.05M | $0.128 | Oct 23, 2026 |
| Kimi K3Moonshot AI | $3 | $15 | $0.30 | 1.05M | $0.192 | Jul 16, 2026 |
| Kimi K2.7 CodeMoonshot AI | $0.95 | $4 | $0.19 | 262K | $0.0616 | Jun 12, 2026 |
| Kimi K2.6Moonshot AI | $0.95 | $4 | $0.16 | 262K | $0.0604 | Apr 20, 2026 |
| Kimi K2.5Cheapest host: SiliconFlow | $0.50 | $2.50 | $0.15 | 262K | $0.036 | Jan 27, 2026 |
| Kimi K2 ThinkingCheapest host: Novita | $0.60 | $2.50 | $0.15 | 262K | $0.04 | Nov 6, 2025 |
| Kimi K2 0905Cheapest host: Novita | $0.60 | $2.50 | $0.60 | 262K | $0.058 | Sep 4, 2025 |
| Kimi K2 0711Cheapest host: Novita | $0.57 | $2.30 | $0.57 | 131K | $0.0548 | Jul 11, 2025 |
Listed is the day OpenRouter added the model. Prices checked Nov 14, 2026.
Kimi comes from Moonshot AI, a lab in Beijing, and it's sold on Moonshot's own platform. Moonshot releases most Kimi models as open weights, so hosts sell them too. This page lists all 8 Kimi models OpenRouter carries, newest first, at Moonshot's own price where Moonshot sells the model.
Kimi K3 is a 2.8 trillion parameter open weight model, at $3 input and $15 output per million tokens. Kimi K3.1 Pro is post-trained from K3, also open weight, with a 1.05M window, at $2 and $10. Kimi K2.7 Code is tuned for end to end programming, at $0.95 and $4.
Calculator
Cost Comparison
Based on 80,000 input tokens (50% cached), 4,000 output tokens, and 100 requests.Prices come live from OpenRouter at each maker's own price where it's listed. Temporary deals are left out.
How to read the Kimi prices
The newer model also costs less here. Kimi K3.1 Pro lists $2 input and $10 output per million, against $3 and $15 on Kimi K3, so the example request runs $0.128 a call against $0.192. Unless your tests favor K3, there's little reason to stay on it.
A token read from the cache costs $0.20 per million on Kimi K3.1 Pro, 90% off the normal input rate. Agent loops that resend a long history every turn get the most out of it. No batch price is listed for Kimi K3.1 Pro right now.
Moonshot publishes Kimi weights under its own license terms, so read them before you ship a self-hosted copy in a large product. Kimi K3.1 Pro reads up to 1.05M tokens, the largest window in the family.
Verdict
Start with Kimi K3.1 Pro and run Kimi K2.6, the lowest cost model Moonshot sells for the example request, on the same prompts. If it passes your checks, use it for volume. For coding agents, try Kimi K2.7 Code too, since that's the job it was tuned for.
Price your own mix of input and output, not just the example. The Kimi models charge input and output in different ratios, so the order can flip with your workload.
More LLM pricing scenarios
Related guides
Frequently asked questions
How much does the Kimi API cost?
Who makes Kimi?
Which Kimi model is best for coding?
Does the Kimi API cache prompts?
Can I run Kimi myself?
Which Kimi model has the largest context window?
Where do these prices come from?
Related tools
LLM VRAM Calculator
Calculate the VRAM needed to run or fine-tune any LLM at any quantization.
Use toolPower Cost Estimator
Estimate annual electricity costs for your PC, Server, or TV.
Use toolData Transfer Calculator
Estimate transfer times for files over USB, WiFi, Ethernet, and more.
Use toolJSON Formatter
Validate, format, and minify JSON. Browse it as a tree, run JSONPath queries, compare two documents, or validate against a JSON Schema.
Use tool