OpenAI

GPT-5.4

Agent-ready GPT for coding and computer-use workflows at a lower cost

openai/gpt-5.4

Context

1.1M

Output

128K

Released

2026-03-05

Knowledge cutoff

2025-08-31

Types

InputTextImagePDF
OutputText

Reasoning

Supported

Tool use

Supported

Structured output

Supported

Attachments

Supported

Open weights

—

Provider pricing

Poe
InTextImagePDF
OutImage
Price
$2.20 / $14.00
Context
1.1M
Status
Stable
Cloudflare AI Gateway
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1M
Status
Stable
CrossModel
InTextImage
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Eden AI
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Impossibl
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Kilo Gateway
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
LLM Gateway
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Merge Gateway
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
NanoGPT
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
NEAR AI Cloud
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Ofox
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
OpenRouter
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
OrcaRouter
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Perplexity Agent
InTextImage
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Prime Inference
InTextImage
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
Vercel AI Gateway
InTextImagePDF
OutText
Price
$2.50 / $15.00
Context
1.1M
Status
Stable
ZenMux
InTextImage
OutText
Price
$3.75 / $18.75
Context
1.1M
Status
Stable
AnyAPI
InTextImagePDF
OutText
Price
—
Context
1.1M
Status
Stable

Benchmarks

  • SWE-Bench Pro

    59.1 · resolve rate

  • SWE-Atlas Codebase QnA

    40.8 · score

  • SWE-Atlas Codebase QnA

    36.3 · score

  • SWE-Atlas Refactoring

    44.29 · score

  • SWE-Atlas Test Writing

    44.36 · score

  • SWE-Atlas Test Writing

    40 · score

  • Artificial Analysis Coding Agent Index

    53.6 · average pass@1

  • SWE-Atlas Codebase QnA

    72.4 · pass@1

  • SWE-Bench Pro

    18.4 · pass@1

  • Terminal-Bench

    69.8 · pass@1

  • Artificial Analysis Coding Agent Index

    52.2 · average pass@1

  • SWE-Atlas Codebase QnA

    72.9 · pass@1

  • SWE-Bench Pro

    18.9 · pass@1

  • Terminal-Bench

    64.7 · pass@1

  • Terminal-Bench

    75.1 · success rate

  • GPQA Diamond

    92.8 · accuracy

  • Humanity's Last Exam

    39.8 · accuracy

  • Humanity's Last Exam

    52.1 · accuracy

  • OSWorld-Verified

    75 · success rate

  • BrowseComp

    82.7 · accuracy

  • GDPval

    83 · wins or ties

  • ARC-AGI-2

    73.3 · accuracy

  • FrontierMath

    47.6 · accuracy

  • FrontierMath

    27.1 · accuracy

  • MMMU Pro

    81.2 · accuracy

From the same lab