GPT-5.2

Dec 2025
GPT-5.2
Compare in
on
—per task?
$1.75 / $14per 1M in / outcache $0.17
95 tok/sfirst token in 88 s?
400Kcontext

Rankingat xhigh effort?

Full ranking →
1Fable 5.199.22Opus 5.597.92GPT-6 Astra97.94Gemini 4 Argon97.14GPT-6.1 Sol97.1
···
27Grok 4.76028GPT-5.259.829Grok 4.558.2

Strong at

S
Long context#22 of 273Instructions#23 of 245
A
Reasoning#53 of 334Science#33 of 312Knowledge#75 of 318
B
Tool calling#64 of 242Vision#54 of 124

Alternatives?

ModelScoreCostPer taskWhat changes
CheaperGemini 4 Argonhigh+62%-17%*$1.99Synchronous output stops at 65,536+2BetterClaude Haiku 5.5xhigh+8%-96%$0.12Remove sampling parameters+5OpenGLM 5.3 Flashmax+12%-95%$0.28Defaults to max · Self-hosted+4CheaperGPT-6.1 Solmax+62%-17%$0.72Remove unsupported parameters+5CheaperGemini 3.8 Flashhigh+55%-69%*$1.24Sampling parameters are ignored+1

History

17 versions
020406080100Jul 24Jan 25Jul 25Jan 26Release dateARC-AGI-2 public evaluation4o4.1 · Apr5 · Aug5.1 · Nov5.2 · Dec5.4 · Mar5.5
02040608010020252026Release dateARC-AGI-2 public evaluation4o4.1 · Apr5 · Aug5.1 · Nov5.2 · Dec5.4 · Mar5.5

Benchmarks

in Chat & Knowledge
AA Intelligence Index 4.3.2≈30.4#55
AA Omniscience Index-0.9#75
AA Omniscience · Accuracy44.3#36
AA Omniscience · Hallucination Rate81.2#171
Humanity's Last Exam37.7#53
GPQA Diamond90.3#33
AA-LCR (long context) 1.182.7#22
IFBench (instructions)75.4#23
CritPt (physics)11.6#53
AIME 2025 · AA archived99#1
ARC-AGI-1 · public evaluation95#24
ARC-AGI-2 · public evaluation59.8#28
ARC-AGI-1 · semi-private86.2#33
ARC-AGI-2 · semi-private52.9#32
MathArena · AIME 2025100#1
MathArena · HMMT February 2025100#1
MathArena · HMMT November 202599.2#1
MathArena · SMT 202596#1
HealthBench Professional length-adjustedlab45.9
MMLU-Pro · official boardlab87.4

FAQ

How much does GPT-5.2 cost?›

$1.75 per 1M input tokens and $14 per 1M output tokens; cached input $0.17 per 1M.

Is GPT-5.2 free to use?›

There is no free API; the cheapest paid price is $1.75 per 1M input tokens.

How good is GPT-5.2, and what is it best at?›

#53 of 334 in Chat & Knowledge, #13 of 19 in Coding, #9 of 16 in Agents & Tool Use, #77 of 195 in Writing & Language, #54 of 124 in Vision & Documents. Strongest at: Reasoning, Science, Long context, Knowledge, Instructions.

Is GPT-5.2 better than Grok 4.7?›

On ARC-AGI-2 · public evaluation, GPT-5.2 scores 59.8 vs 60 — about the same.

How can I use GPT-5.2?›

Through the API of Azure, OpenAI.

What is the context window of GPT-5.2?›

400K tokens, with up to 128K output tokens.

When was GPT-5.2 released? Is there a newer version?›

Released 2025-12-11. It has been superseded by GPT-5.4.

Is GPT-5.2 open source?›

No, GPT-5.2 is a closed model available through an API. The closest open-weights option here is GLM 5.3 Flash.

Can GPT-5.2 read images?›

It accepts text and images, PDF files as input.

What is GPT-5.2?›

GPT-5.2 is a closed language model from OpenAI, released Dec 2025.

What are the best alternatives to GPT-5.2?›

Gemini 4 Argon (cheaper), Claude Haiku 5.5 (better), GLM 5.3 Flash (open weights).