Moonshot AI › Kimi K2.5 › Pricing
$0.60 / $3per 1M tokens in / out
$0.10cache read · −83%
#130on OpenRouter by tokens · 33.2B / week
Price sheetper 1M tokens
| Input | $0.60 | list |
| Output | $3 | |
| Cache read | $0.10 | −83% vs input |
| Context · max output | 256K · 262K | tokens |
Where to buy10 providers · per 1M tokens
| Provider | Input | Output | Context |
|---|---|---|---|
| Azure AI Foundry | $0.11 | $3.30 | — |
| DeepInfra | $0.45 | $2.25 | — |
| AtlasCloud | $0.49 | $2.50 | 262K |
| SiliconFlow | $0.50 | $2.50 | 262K |
| lab-qwen | $0.57 | $3.01 | — |
| Amazon Bedrock | $0.60 | $3 | 262K |
| AWS Bedrock | $0.60 | $3 | — |
| moonshotai | $0.60 | $3 | — |
| Novita | $0.60 | $3 | — |
| Featherless | $0.80 | $3.40 | — |
In Moonshot AI's line-upper 1M in / out
| Kimi K2.5 | $0.60 / $3 | this model |
| Kimi K2.7 Code | $0.95 / $4 | +43% |
| Kimi K2.7 Code Highspeed | $1.90 / $8 | 2.9× |
| Kimi K3.1 Pro | $2 / $10 | 3.3× |
| Kimi K3 | $3 / $15 | 5× |
| Kimi K3.1 Pro Highspeed | $4 / $20 | 6.7× |
Cheaper for the same workvs Kimi K2.5
| MiMo v2.6 Pro thinkingOpen | -55% | score +97% | $0.13 |
| DeepSeek V4.1 Flash 1017 maxOpen | -56% | score +92% | $0.26 |
| Claude Haiku 5.5 max | -83% | score +85% | $0.21 |
| MiniMax M3.1 Flash thinking | -78% | score +82% | $0.22 |
| GLM 5.3 Flash maxOpen | -80% | score +78% | $0.28 |
Pricing questions
How much does Kimi K2.5 cost?›
$0.60 per 1M input tokens and $3 per 1M output tokens. Cached input is $0.10.
Is Kimi K2.5 free to use?›
There is no free API tier, but the weights are open: you can run it on your own hardware for the cost of compute.
How much does prompt caching save?›
Cache reads cost $0.10 per 1M, 83% below the input price.
Is Kimi K2.5 cheaper than Kimi K2 0905?›
No: $0.60 / $3 against $0.40 / $2, 50% more per token.
What is a cheaper alternative to Kimi K2.5?›
MiMo v2.6 Pro at thinking effort costs -55% per task with a score +97% on Humanity's Last Exam.