Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models741Deals30d271Policy30d66, 14 pendingSources1,015

GPT-5.4 mini

OpenAIReleased Mar 17, 2026Current

Score basis

GPT-5.4 mini is capable in long context; and behind the leaders in reasoning, multimodal tasks, agentic tasks, coding, instruction following, and math. Too few results yet to rate safety, multilingual tasks, or factuality.

Capability profile

Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.

Capable1 capability
  1. Long Context−16.4%1 of 3
Limited6 capabilities
  1. Reasoning−27.7%5 of 6
  2. Multimodal−30.6%1 of 6
  3. Agentic−35.3%6 of 7
  4. Coding−35.6%7 of 10
  5. Instruction Following−36.1%2 of 3
  6. Math−52.1%3 of 5
Not rated3 capabilities

Too few results yet to rate Safety, Multilingual or Factuality.

Price

$0.75input$4.50outputper million tokens

From Artificial Analysis · 2 providers tracked · All prices

Costs more than 68% of 330 priced models · 3:1 input-to-output blend, log scale

Evidence

95results on49benchmarks

  • 30 independently verified
  • 54 aggregator
  • 11 vendor-reported

From 13 sources · latest Oct 8, 2026 · How verification works

API features

Tool callingStructured outputsReasoning

As listed by OpenRouter

Research

54 papers reference GPT-5.4 mini

GPT-5.4 mini benchmark results

95 results on 49 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Long Context

CapableFull long context ranking

16.4% behind the leader1 of 3 ranked benchmarks measured

Show 2 more long context resultsHide 2 long context results

Reasoning

LimitedFull reasoning ranking

27.7% behind the leader5 of 6 ranked benchmarks measured

Show 11 more reasoning resultsHide 11 reasoning results

Multimodal

LimitedFull multimodal ranking

30.6% behind the leader1 of 6 ranked benchmarks measured

Show 5 more multimodal resultsHide 5 multimodal results

Agentic

LimitedFull agentic ranking

35.3% behind the leader6 of 7 ranked benchmarks measured

Show 6 more agentic resultsHide 6 agentic results

Coding

LimitedFull coding ranking

35.6% behind the leader7 of 10 ranked benchmarks measured

Show 4 more coding resultsHide 4 coding results

Instruction Following

LimitedFull instruction following ranking

36.1% behind the leader2 of 3 ranked benchmarks measured

Show 2 more instruction following resultsHide 2 instruction following results

Math

LimitedFull math ranking

52.1% behind the leader3 of 5 ranked benchmarks measured

Show 2 more math resultsHide 2 math results

Factuality

Not enough dataFull factuality ranking

0 of 4 ranked benchmarks measured

Show 5 more factuality resultsHide 5 factuality results

More results

Benchmarks outside the capability baskets. They are not ranked against a leader.

Show 25 more resultsHide 25 results

GPT-5.4 mini: common questions

Who makes GPT-5.4 mini?

GPT-5.4 mini is made by OpenAI.

When was GPT-5.4 mini released?

GPT-5.4 mini was released on Mar 17, 2026, according to Artificial Analysis.

What is GPT-5.4 mini good at?

GPT-5.4 mini is capable in long context; and behind the leaders in reasoning, multimodal tasks, agentic tasks, coding, instruction following, and math. Too few results yet to rate safety, multilingual tasks, or factuality.

How much does GPT-5.4 mini cost?

GPT-5.4 mini costs $0.75 per million input tokens and $4.50 per million output tokens, according to Artificial Analysis. We track its price at 2 providers. At a mix of three input tokens to one output token, it costs more than 68% of the 330 priced models we track.

How many benchmarks has GPT-5.4 mini been tested on?

We track 95 results for GPT-5.4 mini on 49 benchmarks from 13 sources, 30 of them independently verified. The latest was recorded on Oct 8, 2026.

Which API features does GPT-5.4 mini support?

OpenRouter lists tool calling, structured outputs, and reasoning for GPT-5.4 mini.

About this record

Where GPT-5.4 mini's numbers come from, and every name it appears under.

Tracked since
Apr 27, 2026
Newest source mention
Sep 9, 2026

Where the results come from

Verification: 95 scores · 30 independently verified · 54 aggregator-attributed · 11 vendor-reported. How these tiers are assigned

From 13 sources on 9 sites. Artificial Analysis supplies 54 of them; the 30 independently verified results come from 7 sites. Bars are coloured by trust tier.

  • artificialanalysis.ai54
  • api.llm-stats.com11
  • arcprize.org8
  • livebench.ai7
  • epoch.ai6
  • raw.githubusercontent.com4
  • datasets-server.huggingface.co2
  • lmarena.ai2
  • labs.scale.com1

Also known as

How our sources name GPT-5.4 mini at each reasoning setting.

SettingShort formAPI id
lowgpt-5.4 mini (low)—
mediumgpt-5.4 mini (medium)gpt-5-4-mini-medium
highgpt-5.4 mini (high)gpt-5.4-mini-high
xhighgpt-5.4 mini (xhigh)—
Also listed asgpt-5.4-mini (Non-Reasoning)gpt-5-4-mini-non-reasoninggpt-5.4-mini (reasoning)gpt-5-4-minigpt-5.4-mini-2026-03-17gpt-5.4-mini:batchgpt-5.4 mini (none)