Skip to content
aiqbench

OpenAI · model passport

OpenAI: o4 Mini High

o4-mini
  • Cloud only
  • Proprietary licence
  • 200,000 context
  • Released 2025-04-16(curated)
20/27axes covered
Overall · LMArena
1,391#161 of 396
Best percentile
97.6LiveBench
Composite standing
55.4shrunk · 20/27 axes

What the data says

Benchmarked on 20 of 27 tracked axes. Every figure below is the model's own published result on that board.

Strongest axes

  • 97.6 pct — LiveBench (78.1%)
  • 96.2 pct — LiveCodeBench (Artificial Analysis) (85.9%)

Weakest axes

  • 7 pct — Factuality Arena (1,368)
  • 9.5 pct — HHEM Hallucination (Vectara) (81.4%)
  • 14.6 pct — SimpleQA Verified (19.6%)

Uneven: 90.6 percentile points between its best and worst axis — 97.6 pct on LiveBench versus 7 pct on Factuality Arena.

  • LiveBench97.6 pct78.1%Verify →
  • LiveCodeBench (Artificial Analysis)96.2 pct85.9%Verify →
  • MMLU-Pro (Artificial Analysis)84.6 pct83.2%Verify →
  • IFBench (Artificial Analysis)79.6 pct68.7%Verify →
  • BFCL v4 (function calling)Dormant board79.3 pct53.2%Verify →
  • SWE-bench Verified (live board)70.8 pct64.6%Verify →
  • FrontierMath65.2 pct24.8%Verify →
  • GPQA Diamond60.6 pct79.6%Verify →
  • Overall59.6 pct1,391Verify →
  • Coding58.1 pct1,433Verify →
  • Hard prompts58.1 pct1,406Verify →
  • AIME (math)58 pct81.7%Verify →
  • Instruction following56.6 pct1,369Verify →
  • Humanity's Last Exam52.2 pct18.1%Verify →
  • Vision Arena47.3 pct1,201Verify →
  • ARC-AGI-239.3 pct6.1%Verify →
  • SimpleBench31 pct38.7%Verify →
  • SimpleQA Verified14.6 pct19.6%Verify →
  • HHEM Hallucination (Vectara)9.5 pct81.4%Verify →
  • Factuality Arena7 pct1,368Verify →
  • WebDev ArenaNot benchmarked
  • Search ArenaNot benchmarked
  • Document ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • Terminal-Bench 4.0Not 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/o4-mini-high
Context length200,000 tokens
Max output100,000 tokens
Acceptsimage, text, file
Input price$1.10 per 1M tokens
Cached input price$0.28 per 1M tokens
Output price$4.40 per 1M tokens

API parameters this endpoint supports

Include reasoningMax tokensReasoningReasoning effortResponse formatSeedStructured outputsTool choiceTool calling

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/o4-mini-high","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.

  • o4-mini-2025-04-16 stands at 55.4 (raw mean 56.3) — measured on 20 of 27 axes.

    standingtemplateVerify →

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 →