kimi-k2.5Moonshot AI
- endpoints
- 7
- regions
- 6
- api
- chat
- released
- March 2026
capabilities 5/8
The same weights cost 1.6x more on aws bedrock than on deepinfra inc.. That is what the endpoints table is for.
| # | flags | ||||||
|---|---|---|---|---|---|---|---|
| 1 |  DeepInfra Inc. | Global | 262K | $0.45 | $2.25 | $0.07 | zdr |
| 2 |  Moonshot AI | Global | 262K | $0.60 | $3.00 | $0.10 | zdr |
| 3 |  AWS Bedrock | us-east-1 | 128K | $0.60 | $3.00 | - | zdr |
| 4 | AWS Bedrock | us-east-2 | 128K | $0.60 | $3.00 | - | zdr |
| 5 | AWS Bedrock | us-west-2 | 128K | $0.60 | $3.00 | - | zdr |
| 6 | AWS Bedrock | eu-north-1 | 128K | $0.72 | $3.60 | $0.72 | zdr |
| 7 | AWS Bedrock | eu-west-2 | 128K | $0.72 | $3.60 | $0.72 | zdr |
no measured traffic for this model yet. the endpoints table above carries provider pricing
| eval | Score as a share of 100 | /100 |
|---|---|---|
| τ²-Bench | 95.9 | |
| GPQA Diamond | 87.9 | |
| Coding Index | 46.8 | |
| Terminal-Bench Hard | 34.8 | |
| Humanity's Last Exam | 30.7 | |
| Intelligence Index | 23.5 | |
| 6 evals | mean | 53.3 |
Scores from Artificial Analysis and public leaderboards, normalised to 100. Bar colour is the band, not the rank: green 80 and up, blue 55 and up, amber 30 and up. Benchmarks measure narrow skills, so test on your own workload before committing. Released 2026-01-27.
deepinfra/moonshotai/Kimi-K2.5this model has no managed policy yet, so call the provider endpoint directly. every id in the endpoints table works the same way
Base url is https://router.requesty.ai/v1 for every id here. One key reaches the whole catalog.
from openai import OpenAI client = OpenAI( api_key="YOUR_REQUESTY_API_KEY", base_url="https://router.requesty.ai/v1",) response = client.chat.completions.create( model="deepinfra/moonshotai/Kimi-K2.5", messages=[ {"role": "user", "content": "Explain quantum computing in one paragraph."}, ],) print(response.choices[0].message.content)
Change the base url, use your Requesty key, set the model to any id in the call pane. Existing OpenAI SDK code needs no other edit, and the same key reaches every other model in the catalog, so switching later is a one-parameter change. Browse all models
Kimi K2.5 is an open-source, native multimodal agentic model built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base. It seamlessly integrates vision and language understanding with advanced agentic capabilities, instant and thinking modes, as well as conversational and agentic paradigms.
questions 5
Which providers serve kimi-k2.5?
How much does kimi-k2.5 cost?
Is kimi-k2.5 open weights?
What is the context window of kimi-k2.5?
How do I use kimi-k2.5 with the OpenAI SDK?
more from moonshot ai 5
| model | context | from /M | providers |
|---|---|---|---|
| kimi-k3 | 1.0M | $3.00 | 5 |
| kimi-k2.7-code | 262K | $0.75 | 4 |
| kimi-k2.6 | 262K | $0.60 | 5 |
| kimi-k2 | 262K | $0.60 | 1 |
| kimi-k2-instruct | 131K | $0.57 | 1 |
route kimi-k2.5 through one endpoint
One key for 3 providers on this model and 600+ others. No markup on provider prices, automatic failover, caching built in. Weekly aggregates in the measured pane come from production traffic routed through Requesty, one line per provider on a shared axis. Methodology