# Kimi K2.5

Kimi K2.5 is a language model from Togethercomputer with a 256K-token context window. Provider list price is $0.50 per million input tokens and $2.80 per million output; on Allocate you pay $0.54 and $3.00. The weights are open, so you can fine-tune it and own the result.

## Pricing

| | Provider list | On Allocate |
| --- | --- | --- |
| Input, per M tokens | $0.50 | $0.54 |
| Output, per M tokens | $2.80 | $3.00 |

Prices checked 2026-07-21.

## Facts

| Field | Value |
| --- | --- |
| Lab | Togethercomputer |
| Modality | Language |
| Context window | 256K tokens |
| License | Not listed |
| Open weights | Yes |
| Fine-tunable | Yes, on your data |
| Catalog id | moonshotai/kimi-k2.5-fp4 |

## What a real workload costs

Take 1,000,000 requests a month at 1,200 input and 350 output tokens each: 1,200M input and 350M output tokens. At list prices that is 1,200 × $0.50 + 350 × $2.80 = $1,580 a month. Billed on Allocate it is $1,691.

## Where it fits

Moonshot’s open-weight model with 256K tokens of context at $0.50 per million input. The open choice for long-document tasks: a policy book, a case history, or a codebase section in one prompt, with weights you can fine-tune.

- Whole-document reasoning
- Long-context retrieval
- Open-weight fine-tuning

## Common questions

### How much does Kimi K2.5 cost per million tokens?

Provider list price is $0.50 per million input tokens and $2.80 per million output tokens. On Allocate you pay $0.54 in and $3.00 out.

### What context window does Kimi K2.5 have?

262,144 tokens (256K). At roughly 0.75 words per token, that is about 197k words of English text per request.

### Can I fine-tune Kimi K2.5?

Yes. Kimi K2.5 is an open-weights model; check the lab’s model card for the exact license terms. Read the license terms before fine-tuning for commercial use. On Allocate the trained weights stay inside your boundary and belong to you.

### How do I call Kimi K2.5 on Allocate?

Send moonshotai/kimi-k2.5-fp4 in the model field of the OpenAI-compatible endpoint at api.allocate.network/v1, or point a route name (like prod/support-agent) at it so you can swap the model later without a deploy.

---

[HTML page](https://allocate.network/models/kimi-k2-5) · [Machine-readable catalog](https://allocate.network/catalog.json)