OpenAI

GPT-5.4 mini

Strong small GPT for coding subagents, quick tool use, and high-volume work

openai/gpt-5.4-mini

Context

400K

Output

128K

Released

2026-03-17

Knowledge cutoff

2025-08-31

Types

InputTextImage
OutputText

Reasoning

Supported

Tool use

Supported

Structured output

Supported

Attachments

Supported

Open weights

—

Provider pricing

Poe
InTextImage
OutText
Price
$0.68 / $4.00
Context
400K
Status
Stable
Cloudflare AI Gateway
InTextImage
OutText
Price
$0.75 / $4.50
Context
128K
Status
Stable
CrossModel
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Eden AI
InTextImagePDF
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
FastRouter
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Impossibl
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Kilo Gateway
InTextImagePDF
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
LLM Gateway
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Merge Gateway
InTextImagePDF
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
NanoGPT
InTextImagePDF
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
NEAR AI Cloud
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Ofox
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
OpenRouter
InTextImagePDF
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
OrcaRouter
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Prime Inference
InTextImage
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
Vercel AI Gateway
InTextImagePDF
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
ZenMux
InText
OutText
Price
$0.75 / $4.50
Context
400K
Status
Stable
AI/ML API
InTextImage
OutText
Price
$2.06 / $12.38
Context
400K
Status
Stable

Benchmarks

  • SWE-Bench Pro

    54.4 · resolve rate

  • Terminal-Bench

    60 · accuracy

  • MCP Atlas

    57.7 · score

  • Toolathlon

    42.9 · score

  • τ²-Bench Telecom

    93.4 · accuracy

  • GPQA Diamond

    88 · accuracy

  • Humanity's Last Exam

    41.5 · accuracy

  • Humanity's Last Exam

    28.2 · accuracy

  • OSWorld-Verified

    72.1 · success rate

  • MMMU Pro

    78 · accuracy

  • MMMU Pro

    76.6 · accuracy

  • OmniDocBench

    0.1263 · overall edit distance

  • OpenAI MRCR

    47.7 · accuracy

  • OpenAI MRCR

    33.6 · accuracy

  • Graphwalks

    76.3 · accuracy

  • Graphwalks

    71.5 · accuracy

From the same lab