> DeepSeek-V3.2 is an open-weights AI model from DeepSeek, released 1 Dec 2025. Index score 145 (GPT-4 = 100), rank 70 of 229 ranked models. API price $0.28 in / $0.42 out per million tokens, 164K context.

[Models](https://themodelindex.org/models/) /[DeepSeek](https://themodelindex.org/labs/deepseek/)

# DeepSeek-V3.2

DeepSeek

Released 1 Dec 2025

Open

[Compare](https://themodelindex.org/compare/deepseek-v3-2-vs-gpt-5-2/) [Announcement](https://api-docs.deepseek.com/news/news251201/)

145134–156

Index score

#70 of 229 models

3/ 8

Core tests taken

3 of 6 areas · plus 11 anchor tests

−16

Versus the frontier at release

best then: Gemini 3 Pro

$0.28/ $0.42

Price per million tokens

input / output, list price

164K

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_

o3 reached this score 8 months earlier.

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

Show 9 anchor tests

Reasoning & knowledge

evidence

GPQA Diamond

83%

Humanity's Last Exam

no result

Maths

evidence

FrontierMath T1-3

no result

Coding

evidence

SWE-bench Verified

no result

Terminal-Bench

40%

Agents

evidence

APEX-Agents

no result

Novel problems

evidence

ARC-AGI-2

4%

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

#70

of 219 · 145

LMArena · Creative writing

creative writing

#95

of 386 · 1401

LMArena · Maths

maths questions in chat

#97

of 373 · 1428

LMArena · Coding

coding questions in chat

#100

of 383 · 1476

LMArena · Text

overall chat quality, judged by people

#107

of 388 · 1425

Epoch Capabilities Index

overall capability from many benchmarks

#70

of 239 · 146.3

LMArena · WebDev

building web apps and front ends

#89

of 123 · 1362

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

Not in the top three among current models on any category we track.

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

DeepSeek Chat

app · subscription

↗

DeepSeek API

API

↗

#### API providers · 13 · per million tokens, in / out

G

GMICloud

$0.21 / $0.31 · fp8

cheapest

D

DeepInfra

$0.26 / $0.38 · fp4

↗

A

AtlasCloud

$0.26 / $0.38 · fp8

↗

V

Venice

$0.27 / $0.39

↗

S

SiliconFlow

$0.26 / $0.42 · fp8

↗

B

Baidu

$0.28 / $0.42 · fp8

↗

D

DigitalOcean

$0.30 / $0.96

↗

A

Alibaba

$0.37 / $1.11 · fp8

↗

#### Run it yourself

🤗

Official weights

deepseek-ai/DeepSeek-V3.2

↗

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

1 Dec 2025

Announcement

[api-docs.deepseek.com ↗](https://api-docs.deepseek.com/news/news251201/)

Release entry

DeepSeek-V3.2 (+ Speciale)

Scored as

DeepSeek-V3.2

Artificial Analysis index

–

Output speed

–

Hugging Face

[deepseek-ai/DeepSeek-V3.2 ↗](https://huggingface.co/deepseek-ai/DeepSeek-V3.2)

## DeepSeek releases · around this one

DeepSeek-V3.2-Exp

Open

29 SEP 2025

index 143

DeepSeekMath-V2

Open

27 NOV 2025

DeepSeek-V3.2

Open

1 DEC 2025

index 145

DeepSeek-V4-Flash (Apr 2026)

Open

24 APR 2026

index 145

DeepSeek-V4-Pro (Apr 2026)

Open

24 APR 2026

index 151

---
Source: https://themodelindex.org/models/deepseek-v3-2/ · The Model Index · data as of 2026-10-01