Skip to content
aiqbench

OpenAI · model passport

OpenAI: GPT-5.4 Nano

gpt-5-4-nano
  • Cloud only
  • Proprietary licence
  • 400,000 context
  • Released 2026-10-09(first seen on LMArena)
14/27axes covered
Overall · LMArena
1,401#148 of 396
Best percentile
99.0HHEM Hallucination (Vectara)
Cohort standing
58.2median 60.7

What the data says

Benchmarked on 14 of 27 tracked axes. Every figure below is the model's own published result on that board.

Strongest axes

  • 99 pct — HHEM Hallucination (Vectara) (96.9%)
  • 93.5 pct — IFBench (Artificial Analysis) (75.9%)

Weakest axes

  • 6.1 pct — SimpleQA Verified (11.7%)

Uneven: 92.9 percentile points between its best and worst axis — 99 pct on HHEM Hallucination (Vectara) versus 6.1 pct on SimpleQA Verified.

  • HHEM Hallucination (Vectara)99 pct96.9%Verify →
  • IFBench (Artificial Analysis)93.5 pct75.9%Verify →
  • LiveBench81.1 pct69.6%Verify →
  • Coding68 pct1,460Verify →
  • AIME (math)67.5 pct87.8%Verify →
  • FrontierMath66.7 pct25.9%Verify →
  • Hard prompts64.1 pct1,421Verify →
  • Overall62.9 pct1,401Verify →
  • Instruction following62.4 pct1,387Verify →
  • GPQA Diamond59.3 pct78.5%Verify →
  • Vision Arena45.3 pct1,199Verify →
  • ARC-AGI-237.1 pct5.7%Verify →
  • Factuality Arena26.7 pct1,417Verify →
  • SimpleQA Verified6.1 pct11.7%Verify →
  • WebDev ArenaNot benchmarked
  • Search ArenaNot benchmarked
  • Document ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • Humanity's Last ExamNot benchmarked
  • SimpleBenchNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • Terminal-Bench 4.0Not benchmarked
  • MMLU-Pro (Artificial Analysis)Not benchmarked
  • LiveCodeBench (Artificial Analysis)Not benchmarked
  • SWE-bench VerifiedNot benchmarked
  • Terminal-BenchNot benchmarked

Percentiles are the model's standing within each board's own population. Axes marked "not benchmarked" have no published result for this model — that is an absence of data, not a low score.

How to run it

Use it hosted

API model IDopenai/gpt-5.4-nano
Context length400,000 tokens
Max output128,000 tokens
Acceptsfile, image, text
Input price$0.20 per 1M tokens
Cached input price$0.02 per 1M tokens
Output price$1.25 per 1M tokens

API parameters this endpoint supports

Include reasoningMax completion tokensMax tokensReasoningReasoning effortResponse formatSeedStructured outputsTool choiceTool callingVerbosity

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.4-nano","messages":[{"role":"user","content":"Hello"}]}'

View on OpenRouter →

Identifiers, limits and prices are synced from the upstream source shown above. Any field we haven't synced is left out entirely rather than guessed.

  • gpt-5.4-nano-high stands at 58.2 (raw mean 60) — measured on 14 of 27 axes.

    standingtemplateVerify →

  • gpt-5.4-nano-high stands at 58.2 versus a cohort median of 60.7 across 171 models released within 180 days — below the peer median.

    cohorttemplateVerify →

Every claim is a computed fact from the current board data, phrased by a low-temperature model call and rejected by a validation gate if any number is not in the source row. How the leaderboards work →