GPT-4.1

Apr 2025
Retires Jan 14, 2027Move toGPT-6 LunaNewer version
GPT-4.1
Compare in
on
—per task?
$2 / $8per 1M in / outcache $0.50
135 tok/sfirst token in 1.1 s?
1Mcontext

Rankingin non-thinking mode?

Strong at

B
Tool calling#113 of 242Long context#133 of 273
C
Reasoning#302 of 334Science#205 of 312Knowledge#173 of 318Instructions#141 of 245Vision#94 of 124

Alternatives?

ModelScoreCostPer taskWhat changes
CheaperMuse Spark 1.3max×3.8-43%$1.60Only auto tool_choice is supported+5BetterGPT-6 Lunalow+70%-94%$0.005Function calling only with none · Chat Completions+3OpenMiMo v2.6 Prothinking×3.7-84%$0.13Pass back reasoning_content+3CheaperGrok 4.7xhigh×3.7-14%$3.74Context: 1M → 500K+5CheaperQwen3.8 Max (0902)thinking×3.6-14%$5.41Context: 1M → 984K

History

17 versions
10203040Jan 24Jul 24Jan 25Jul 25Jan 26Release dateAA Intelligence Index 4.3.244o4o (2024-05-134o (2024-08-064.1 · Apr5 · Aug5 · Aug5.1 · Nov5.2 · Dec5.4 · Mar5.5
10203040202420252026Release dateAA Intelligence Index 4.3.244o4o (2024-05-134.1 · Apr5 · Aug5 · Aug5.1 · Nov5.2 · Dec5.4 · Mar5.5

Benchmarks

all categories
Chat & Knowledge
Humanity's Last Exam4.2#305
AA Omniscience Index-39.6#174
CritPt (physics)0#180
AA Intelligence Index 4.3.2≈12.7#173
AA Omniscience · Accuracy27.8#89
AA Omniscience · Hallucination Rate93.3#289
AA-LCR (long context) 1.168.3#133
MMLU-Pro · official boardlab81.8
GPQA Diamond66.6#207
AIME 2025 · AA archived34.7#102
Coding
Terminal-Bench Hard (version unstated)13.6#114
LiveCodeBench · AA archived45.7#98
Agents & Tool Use
τ²-bench · AA47.1#115
Writing & Language
IFBench (instructions)43#142
COMET22 averagelab84.1
COMET22 multi-domain Zh-Enlab86.9
COMET22 multi-domain En-Zhlab82.8
COMET22 multi-domain En-Delab82.7
COMET22 multi-domain De-Enlab83.9
COMET22 WMT24lab85.1
Vision & Documents
MMMU-Pro (vision)61.2#96

FAQ

How much does GPT-4.1 cost?›

$2 per 1M input tokens and $8 per 1M output tokens; cached input $0.50 per 1M.

Is GPT-4.1 free to use?›

There is no free API; the cheapest paid price is $2 per 1M input tokens.

How good is GPT-4.1, and what is it best at?›

#302 of 334 in Chat & Knowledge, #113 of 239 in Coding, #113 of 242 in Agents & Tool Use, #65 of 195 in Writing & Language, #94 of 124 in Vision & Documents.

Is GPT-4.1 better than Qwen3 235B A22B Thinking 2507?›

On AA Intelligence Index 4.3.2, GPT-4.1 scores 12.7 vs 12.7 — about the same.

How can I use GPT-4.1?›

Through the API of Azure, Azure AI Foundry, OpenAI.

What is the context window of GPT-4.1?›

1M tokens, with up to 33K output tokens.

When was GPT-4.1 released? Is there a newer version?›

Released 2025-04-14. It has been superseded by GPT-6 Luna.

Is GPT-4.1 open source?›

No, GPT-4.1 is a closed model available through an API. The closest open-weights option here is MiMo v2.6 Pro.

Can GPT-4.1 read images?›

It accepts text and images, PDF files as input.

What is GPT-4.1?›

GPT-4.1 is a closed language model from OpenAI, released Apr 2025, the successor to GPT-4.5 Preview.

What are the best alternatives to GPT-4.1?›

Muse Spark 1.3 (cheaper), GPT-6 Luna (better), MiMo v2.6 Pro (open weights).