> Kimi K2.5 is an open-weights AI model from Moonshot, released 27 Jan 2026. Index score 149 (GPT-4 = 100), rank 60 of 229 ranked models. API price $0.45 in / $2.25 out per million tokens, 262K context.

[Models](https://themodelindex.org/models/) /[Moonshot](https://themodelindex.org/labs/moonshot/)

# Kimi K2.5

Moonshot

Released 27 Jan 2026

Open

[Compare](https://themodelindex.org/compare/kimi-k2-5-vs-gpt-5-3-codex/) [Announcement](https://www.kimi.com/blog/kimi-k2-5?ref=upstract.com)

149138–161

Index score

#60 of 229 models

5/ 8

Core tests taken

3 of 6 areas · plus 12 anchor tests

−13

Versus the frontier at release

best then: GPT-5.2

Replaced by [Kimi K2.6](https://themodelindex.org/models/kimi-k2-6/)

$0.45/ $2.25

Price per million tokens

input / output, list price

262K

Context window

tokens it reads at once

## Where it sits · on the frontier

_Chart: Model Index score of every scored model since late 2022, with the best closed and best open model over time_

GPT-5 reached this score 6 months earlier.

## Test results · by area · tick = best by any model

Show 9 anchor tests

Reasoning & knowledge

evidence

GPQA Diamond

88%

Humanity's Last Exam

24%

Maths

evidence

FrontierMath T1-3

no result

Coding

evidence

SWE-bench Verified

74%

Terminal-Bench

43%

Agents

evidence

APEX-Agents

no result

Novel problems

evidence

ARC-AGI-2

12%

Long tasks

no evidence yet

METR time horizon

no result

Independent results from the Epoch AI benchmarking hub, best reasoning setting. The core tests are shown here; anchor tests also inform the score and can be shown above.

## What others say · rank on each leaderboard

The Model Index

independent tests, our method

#60

of 219 · 149

LMArena · Maths

maths questions in chat

#45

of 373 · 1471

LMArena · Coding

coding questions in chat

#65

of 383 · 1502

LMArena · Text

overall chat quality, judged by people

#68

of 388 · 1450

LMArena · Creative writing

creative writing

#69

of 386 · 1423

Epoch Capabilities Index

overall capability from many benchmarks

#58

of 239 · 148.0

LMArena · Vision

understanding images

#42

of 149 · 1252

LMArena · WebDev

building web apps and front ends

#66

of 123 · 1436

LMArena · Documents

working with long documents

#32

of 41 · 1430

Further left is better. Leaderboards measure different things: LMArena is people voting blind between two answers, the others are test scores. Click any row for the source.

## Best at · top three among current models

3rd

SWE-bench Verified

Epoch AI · independent

Where this model places in the top three of today's models, on category leaderboards and on our core tests.

## Where to use it · prices checked 1 Oct 2026

#### From the lab

Kimi

app · subscription

↗

Moonshot API

API

↗

#### API providers · 5 · per million tokens, in / out

S

SiliconFlow

$0.45 / $2.25 · int4

cheapest

A

AtlasCloud

$0.49 / $2.5 · int4

↗

N

Novita

$0.57 / $2.85

↗

A

Amazon Bedrock

$0.60 / $3.0

↗

V

Venice

$0.53 / $3.33

↗

#### Run it yourself

🤗

Official weights

moonshotai/Kimi-K2.5

↗

Provider prices come from OpenRouter's live listing; going direct to a provider can differ. Subscriptions are the lab's own apps.

## Facts and sources

Released

27 Jan 2026

Announcement

[kimi.com ↗](https://www.kimi.com/blog/kimi-k2-5?ref=upstract.com)

Release entry

Kimi K2.5

Scored as

Kimi K2.5

Artificial Analysis index

–

Output speed

–

Hugging Face

[moonshotai/Kimi-K2.5 ↗](https://huggingface.co/moonshotai/Kimi-K2.5)

## Moonshot releases · around this one

Kimi Linear 48B-A3B

Open

30 OCT 2025

Kimi K2 Thinking

Open

6 NOV 2025

index 144

Kimi K2.5

Open

27 JAN 2026

index 149

Kimi K2.6

Open

20 APR 2026

index 156

Kimi K2.7 Code

Open

12 JUN 2026

index 153

---
Source: https://themodelindex.org/models/kimi-k2-5/ · The Model Index · data as of 2026-10-01