BenchLeader

o4-mini

Reasoning effort
Quality#80 of 333
56.9Index
Top quarter
+6.9 on the median
Price#169 of 249
$1.93per 1M tokens
Dearer than most
2.3× the median

20 cheaper ranked models score higher: Gemini 3.7 Flash (62.2, $1.50), Gemini 3.8 Flash (62.0, $1.50), Gemini 3.6 Flash (61.3, $1.50) and 17 more.

As of 8 Nov 2026, o4-mini by OpenAI scores 56.9 on the BenchLeader Index (593 ranked configurations), with its best results in maths (69) and weakest in knowledge (38). API pricing: $1.10 per 1M input tokens and $4.40 per 1M output tokens ($1.93 blended at 3:1), list price on 8 Nov 2026. Released 16 Apr 2025.

Key facts (as of 8 Nov 2026)

  • o4-mini API pricing: $1.10 per 1M input tokens and $4.40 per 1M output tokens ($1.93 blended at 3:1) (list price, 8 Nov 2026).
  • o4-mini was released on 16 Apr 2025 (OpenAI).

In detail. o4-mini is an OpenAI proprietary reasoning model, released 16 Apr 2025. Its best configuration ranks #170 of 593 on the BenchLeader Index at 56.9 ±5.9, in the upper half. The ± is the point: 293 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in maths (69) and lowest in knowledge (38). At $1.93 per million tokens blended it is pricier than most ranked models. It has been measured at 4 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 6 Nov 2026.

Blended price
$1.93/M
$1.10 in · $4.40 out
via models.dev
Output speed
–
First answer
–
Context
200k
via models.dev
Full answer
–
Cost per task
–
Tokens per task
–
Released
16 Apr 2025
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index57
  2. Reasoning59
  3. Coding54
  4. Agents & tools64
  5. Maths69
  6. Knowledge38
  7. Instruction following69
  8. Human preference56
  9. Multimodal57

Get an email when OpenAI ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

EffortIndexRankSpeedFirst answerChat reply costCategories
low41.4#500––$0.0018Knowledge 32 · Maths 39 · Reasoning 41
medium51.8#296––$0.0018Long context 54 · Maths 46 · Multimodal 66 · Reasoning 44
high50.3#33292 tok/s27 s$0.0018Agents & tools 43 · Coding 52 · Instruction following 50 · Knowledge 47 · Long context 55 · Maths 51 · Multimodal 59 · Reasoning 46
not statedbest56.9#170––$0.0018Agents & tools 64 · Coding 54 · Human preference 56 · Instruction following 69 · Knowledge 38 · Maths 69 · Multimodal 57 · Reasoning 59

How its index has moved

8 Sep 2026 to 8 Nov 2026
365064
  • not stated
  • medium
  • high
  • low

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmarklowmediumhighnot statedSourceTrend
GPQA Diamond75.3%#16077.8%#14479.6%#137–Epoch AI Benchmarking Hub
Humanity's Last Exam–14.3%#2618.1%#25–Scale AI / CAIS
SimpleBench––38.7%#62–SimpleBench
Arena Hard Promptseffort not stated–––1406#174Arena (LMArena)
GPQA Diamond (AA)not in index––78.4%#213–Artificial Analysis
Humanity's Last Exam (AA)not in index––16.5%#252–Artificial Analysis
GPQA Diamond (Vals)not in index––74.5%#82–Vals AI
Kagi LLM Benchmarkeffort not stated–––67.6%#32Kagi LLM Benchmark
GPQA Diamond (HELM)not in indexeffort not stated–––73.5%#6HELM Capabilities
ARC-AGI-121.3%#22241.8%#19258.7%#165–ARC Prize
ARC-AGI-21.7%#2132.4%#2036.1%#173–ARC Prize
CritPtnot in index––0.6%#266–Artificial Analysis
SciPredictnot in indexeffort not stated–––16.2%#14Scale AI SEAL
Chess Puzzlesnot in index15.0%#11822.0%#8913.0%#124–Epoch AI Benchmarking Hub
Mystery Game Puzzlesnot in index––5.0%#135–Epoch AI Benchmarking Hub
LMCAnot in index––26.5%#148–Epoch AI Benchmarking Hub
DTBenchnot in index––77.6%#143–Epoch AI Benchmarking Hub
ForecastBenchnot in indexeffort not stated–––64.2%#32ForecastBench

Coding

Benchmarkmediumhighnot statedSourceTrend
WeirdML–52.6%#64–WeirdML
GSO-Bench–3.6%#31–GSO-Bench
Arena Codingeffort not stated––1433#172Arena (LMArena)
LiveCodeBench–82.2%#54–Vals AI
Aider Polygloteffort not stated––72.0%#8Aider polyglot leaderboard
SWE-bench Verified (bash only)effort not stated––45.0%#35SWE-bench
SWE-bench Verified (any scaffold)not in indexeffort not stated––64.6%#23SWE-bench
CadEvalnot in index62.0%#3––Epoch AI Benchmarking Hub
ALE-Benchnot in index–826.2#74–Epoch AI Benchmarking Hub
AlgoTunenot in index–1.7#6–AlgoTune
Design Arena Websitenot in indexeffort not stated––983#169Design Arena
LiveCodeBench (AA)not in index–85.9%#12–Artificial Analysis

Agents & tools

Benchmarkmediumhighnot statedSourceTrend
GDPval–25.3%#8–OpenAI
Terminal-Bench Hard–15.2%#197–Artificial Analysis
τ²-Bench Telecom (AA)not in index–55.6%#188–Artificial Analysis
BFCL Overalleffort not stated––53.2%#19Berkeley Function Calling Leaderboard
METR Time Horizonsnot in index63.9%#16––METR
BrowseCompnot in indexeffort not stated––3.3%#7OpenAI

Maths

Benchmarklowmediumhighnot statedSourceTrend
FrontierMath Tiers 1–316.1%#10428.8%#7936.1%#68–Epoch AI Benchmarking Hub
FrontierMath Tier 4––4.9%#68–Epoch AI Benchmarking Hub
OTIS Mock AIME57.8%#17773.3%#13881.7%#121–Epoch AI Benchmarking Hub
MATH Level 5––97.8%#4–Epoch AI Benchmarking Hub
AIME (Vals)––83.7%#41–Vals AI
MGSM––93.4%#9–Vals AI
IMO 2025––14.3%#5–MathArena
Omni-MATH (HELM)effort not stated–––72.0%#2HELM Capabilities
MATH 500not in index––94.2%#10–Vals AI
Arena Mathseffort not stated–––1418#128Arena (LMArena)

Knowledge

Benchmarklowhighnot statedSourceTrend
SimpleQA Verified19.6%#8418.8%#86–Epoch AI Benchmarking Hub
MMLU-Pro–80.6%#85–Vals AI
LegalBench–79.2%#96–Vals AI
CorpFin–59.0%#79–Vals AI
TaxEval–74.8%#29–Vals AI
MedQA–96.0%#9–Vals AI
PRBench Financeeffort not stated––39.2%#29Scale AI SEAL
PRBench Legaleffort not stated––38.1%#38Scale AI SEAL
MultiNRC–22.2%#33–Scale AI SEAL
MMLU-Pro (HELM)not in indexeffort not stated––82.0%#13HELM Capabilities
MedCodenot in index–33.8%#83–Vals AI
MedScribenot in index–69.1%#96–Vals AI

Instruction following

Benchmarkhighnot statedSourceTrend
IFBench68.7%#79–Artificial Analysis
MultiChallenge44.9%#32–Scale AI SEAL
IFEval (HELM)effort not stated–92.9%#5HELM Capabilities
WildBench (HELM)effort not stated–85.4%#9HELM Capabilities
Arena Instruction Followingnot in indexeffort not stated–1368#180Arena (LMArena)

Human preference

Benchmarkmediumnot statedSourceTrend
Arena Texteffort not stated–1391#166Arena (LMArena)
Arena Creative Writingnot in indexeffort not stated–1337#185Arena (LMArena)
Arena Multi-turnnot in indexeffort not stated–1383#178Arena (LMArena)
Arena Longer Queriesnot in indexeffort not stated–1370#194Arena (LMArena)
Lech Mazur Creative Writingnot in index75.0%#27–Lech Mazur

Multimodal

Benchmarkmediumhighnot statedSourceTrend
Arena Visioneffort not stated––1194#96Arena (LMArena)
MMMU-Pro–69.3%#164–Artificial Analysis
VISTA51.7%#551.8%#4–Scale AI SEAL
MMMU (validation)effort not stated––81.6%#4MMMU
MMMU-Pro (Vals)not in index–79.7%#46–Vals AI
MortgageTaxnot in index–64.8%#45–Vals AI
VTBnot in indexeffort not stated––11.1%#10Scale AI SEAL
GeoBenchnot in index64.0%#1464.0%#14–GeoBench
VPCTnot in index57.5%#8––Epoch AI Benchmarking Hub

Long context

BenchmarkmediumhighSourceTrend
Fiction.LiveBench 120k62.5%#13–Fiction.live
AA-LCR v1.1–61.0%#289Artificial Analysis

Composite

Benchmarkhighnot statedSourceTrend
Epoch Capabilities Indexnot in indexeffort not stated–145.2#87Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.2not in index16.6#280–Artificial Analysis
AA Progress Index v1not in index247.9#269–Artificial Analysis
HELM Capabilities meannot in indexeffort not stated–81.2%#2HELM Capabilities

Safety & honesty

Benchmarkmediumhighnot statedSourceTrend
FORTRESSnot in indexeffort not stated––21.5%#42Scale AI SEAL
MASKnot in indexpossible contamination72.9%#2878.6%#27–Scale AI SEAL
PropensityBenchnot in indexeffort not stated––15.8%#12Scale AI SEAL

Where it wins

Benchmarks where this configuration ranks in the top five of every configuration measured.

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.275 per 1M. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0018$0.0015–
Summarize a 30-page report12,000 / 600$0.016$0.0084–
Code edit6,000 / 1,500$0.013$0.0095–
Agentic coding session60,000 / 4,000$0.084$0.046–
Structured extraction2,000 / 200$0.0031$0.0018–

More from OpenAI

New this month

See also

Embed or cite

Show where o4-mini ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page. To see ranks like this on OpenRouter and Hugging Face model pages, get the Chrome extension.

[![o4-mini on BenchLeader: #80 overall](https://www.benchleader.com/badge/o4-mini.svg)](https://www.benchleader.com/models/o4-mini)

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 8). o4-mini on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 8, 2026, from https://www.benchleader.com/models/o4-mini

BibTeX:

@misc{benchleader_o4_mini_20261108,
  author       = {{BenchLeader}},
  title        = {{o4-mini on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/o4-mini}},
  note         = {Data as of 2026-11-08}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are o4-mini's benchmark scores?
As of 8 Nov 2026, o4-mini scores 56.9 ±5.9 on the BenchLeader Index, strongest in maths (69) and weakest in knowledge (38). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does o4-mini rank?
o4-mini is #170 of 593 ranked configurations on 8 Nov 2026. 293 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 69.4.
How much does o4-mini cost per 1M tokens?
As of 8 Nov 2026, o4-mini costs $1.10 per 1M input tokens and $4.40 per 1M output tokens, $1.93 blended (three input tokens per output token).
When was o4-mini released?
OpenAI released o4-mini on 16 Apr 2025. BenchLeader has results for it at 4 reasoning-effort settings.

Data as of 8 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “o4-mini: benchmarks, pricing, speed and rank”, data as of 8 Nov 2026, https://www.benchleader.com/models/o4-mini.