Noometry

OpenAI, proprietary

GPT-5 Mini

GPT-5 Mini by OpenAI ranks 128th of 354 ranked models on the Noometry Index as of October 2026, with a score of 41.8. Its strongest category is instruction following, where it ranks 46th. API pricing starts at $0.25 per million input tokens and $2 per million output tokens, with a 400K-token context window.

Last verified

Noometry rank
#128 of 354
Index score
41.8
Evidence
Confirmed 60 results
Provider
OpenAI
Released
August 7, 2025
Weights
Proprietary
Reasoning
Yes
Context window
400K
Max output
128K
Input price
$0.25 / M
Output price
$2 / M
Blended price
$0.69 / M
Output speed
3 tokens/s Kagi
Value
#90 of 219
Knowledge cutoff
May 2024
Input
text, image

Category scores

Each category score combines every public result we have in that category.

  1. Coding 40.1
  2. Agentic & Tool Use 31.1
  3. Reasoning 23.9
  4. Math 46.7
  5. Knowledge 45.6
  6. Multimodal 35.6
  7. Multilingual 48.9
  8. Instruction Following 76.2
  9. Long Context 41.9
  10. Writing & Preference 55.2
CategoryScoreRankResults
Coding40.1#1466
Agentic & Tool Use31.1#702
Reasoning23.9#16810
Math46.7#697
Knowledge45.6#868
Multimodal35.6#852
Multilingual48.9#1371
Instruction Following76.2#462
Long Context41.9#1322
Writing & Preference55.2#1486

Strengths and weaknesses

Categories where GPT-5 Mini places highest and lowest among the models ranked in each, with its score against that category's median.

Strongest categories

CategoryScorevs medianRank
Instruction Following76.2+5.0#46 of 305, top 16%
Math46.7+10.1#69 of 327, top 22%
Knowledge45.6+8.3#86 of 314, top 28%

Weakest categories

CategoryScorevs medianRank
Multimodal35.6−2.9#85 of 128, top 67%
Reasoning23.9+0.3#168 of 350, top 48%
Writing & Preference55.2+1.4#148 of 312, top 48%

Closest competitors

The models ranked just above and below GPT-5 Mini. When scores are this close, price and speed are often the better way to choose.

ModelRankScoreBlended $/MSpeed
GLM-4.7#12442.0$1—Compare
GPT-5.4 nano#12541.9$0.4619Compare
Amazon Nova Experimental Chat 10 09#12641.9——Compare
Qwen3.5 27B#12741.9$0.82—Compare
ERNIE 5.0 0110#12941.8——Compare
Granite 4.2 30b#13041.8——Compare
Muse Glimmer#13141.7——Compare
o4-mini#13241.6$1.936Compare

Sponsored placements are available on pages like this one. Advertise on Noometry

Benchmark results

Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.

Coding

BenchmarkScorePositionSettingSourceDate
SWE-bench Verified64.7%#26 of 32, top 82%mediumEpoch AI2026-02-01
SWE-bench Verified (bash only)59.8%#20 of 39, top 52%mediumSWE-bench2025-08-07
SWE-bench Multilingual39.7%#13 of 13, top 100%SWE-bench2026-02-13
SciCode39.2%#83 of 121, top 69%Epoch AI
SciCode39%highEpoch AI
WeirdML52.7%#42 of 119, top 36%highEpoch AI
LMArena Coding1406#133 of 294, top 46%highLMArena2026-10-08
ALE-Bench799.77#57 of 105, top 55%highEpoch AI
AlgoTune1.38#15 of 18, top 84%highEpoch AI

Agentic & Tool Use

BenchmarkScorePositionSettingSourceDate
Terminal-Bench34.8%#29 of 41, top 71%Epoch AI
Terminal-Bench31.9%mediumEpoch AI
Berkeley Function Calling Leaderboard55.5%#14 of 49, top 29%fcBerkeley Function Calling Leaderboard
Vending-Bench 2-31.18#60 of 60, top 100%Epoch AI

Reasoning

BenchmarkScorePositionSettingSourceDate
ARC-AGI-24.4%#60 of 83, top 73%highEpoch AI
ARC-AGI-20.8%lowEpoch AI
ARC-AGI-24%mediumEpoch AI
ARC-AGI-21.7%minimalEpoch AI
Kagi LLM Benchmark70.3%#24 of 99, top 25%Kagi LLM Benchmark
ARC-AGI-154.3%#54 of 83, top 66%highEpoch AI
ARC-AGI-126.3%lowEpoch AI
ARC-AGI-137.3%mediumEpoch AI
ARC-AGI-15.3%minimalEpoch AI
CritPt0%#114 of 134, top 86%Epoch AI
CritPt0%highEpoch AI
Chess Puzzles30%#37 of 129, top 29%highEpoch AI2026-08-07
Chess Puzzles12%lowEpoch AI2026-08-07
Chess Puzzles7%minimalEpoch AI2026-08-07
EnigmaEval8.2%#16 of 38, top 43%Epoch AI
LMArena Hard Prompts1380#138 of 297, top 47%highLMArena2026-10-08
Mystery Game Puzzles5%highEpoch AI2026-08-27
Mystery Game Puzzles4%mediumEpoch AI2026-08-27
Mystery Game Puzzles10%#60 of 74, top 82%minimalEpoch AI2026-08-27
DTBench80.5%#71 of 151, top 48%highEpoch AI
LMCA34.2%#65 of 125, top 52%highEpoch AI
Epoch Capabilities Index145.52#81 of 213, top 39%Epoch AI2025-08-07
ForecastBench61#19 of 72, top 27%Epoch AI

Math

BenchmarkScorePositionSettingSourceDate
FrontierMath (Tiers 1-3)46.7%#48 of 81, top 60%highEpoch AI2026-06-12
FrontierMath (Tiers 1-3)18.2%lowEpoch AI2026-08-27
FrontierMath (Tiers 1-3)6%minimalEpoch AI2026-08-27
FrontierMath Tier 412.2%#51 of 63, top 81%highEpoch AI2026-06-12
OTIS Mock AIME 2024-202586.7%#61 of 173, top 36%highEpoch AI2025-10-30
OTIS Mock AIME 2024-202578.3%mediumEpoch AI2025-08-07
OTIS Mock AIME 2024-202555.6%minimalEpoch AI2026-08-07
ProofBench9%#64 of 77, top 84%highEpoch AI
Omni-MATH72.2%Best of 57HELM Capabilities
LMArena Math1378#144 of 285, top 51%highLMArena2026-10-08
MATH Level 597.8%#2 of 79, top 3%highEpoch AI2025-10-30
MATH Level 596.8%mediumEpoch AI2025-08-20
FrontierMath (Feb 2025 set)27.2%#22 of 68, top 33%highEpoch AI2025-11-13
FrontierMath (Feb 2025 set)20.3%mediumEpoch AI2025-11-13
FrontierMath Tier 4 (v1)6.3%#24 of 55, top 44%highEpoch AI2025-10-30
FrontierMath Tier 4 (v1)4.2%mediumEpoch AI2025-08-07

Knowledge

BenchmarkScorePositionSettingSourceDate
GPQA Diamond75%#94 of 186, top 51%highEpoch AI2025-10-30
GPQA Diamond71.7%mediumEpoch AI2025-08-07
GPQA Diamond71.7%minimalEpoch AI2026-08-07
Humanity's Last Exam19.4%#19 of 41, top 47%Epoch AI
SimpleQA Verified21.6%#63 of 77, top 82%highEpoch AI2026-08-10
MMLU-Pro83.5%#11 of 58, top 19%HELM Capabilities
Confabulations (lower is better)13.3%#11 of 51, top 22%medium reasoningLech Mazur benchmarks
Vectara Hallucination Rate (lower is better)12.9%#79 of 96, top 83%Vectara Hallucination Leaderboard
GPQA (HELM)75.6%#3 of 57, top 6%HELM Capabilities
LMArena Expert1379#137 of 273, top 51%highLMArena2026-10-08

Multimodal

BenchmarkScorePositionSettingSourceDate
LMArena Vision1202#77 of 122, top 64%highLMArena2026-10-09
VPCT39%highEpoch AI
VPCT40.2%#11 of 24, top 46%mediumEpoch AI

Multilingual

BenchmarkScorePositionSettingSourceDate
LMArena Non-English1363#137 of 297, top 47%highLMArena2026-10-08
LMArena Chinese1385#146 of 285, top 52%highLMArena2026-10-08
LMArena French1386#128 of 223, top 58%highLMArena2026-10-08
LMArena German1366#120 of 231, top 52%highLMArena2026-10-08
LMArena Japanese1341#105 of 211, top 50%highLMArena2026-10-08
LMArena Korean1308#125 of 213, top 59%highLMArena2026-10-08
LMArena Russian1362#140 of 283, top 50%highLMArena2026-10-08
LMArena Spanish1355#140 of 226, top 62%highLMArena2026-10-08

Instruction Following

BenchmarkScorePositionSettingSourceDate
IFEval92.7%#6 of 57, top 11%HELM Capabilities
LMArena Instruction Following1357#139 of 298, top 47%highLMArena2026-10-08

Long Context

BenchmarkScorePositionSettingSourceDate
Fiction.LiveBench69.4%#17 of 47, top 37%mediumEpoch AI
LMArena Longer Query1355#149 of 291, top 52%highLMArena2026-10-08

Writing & Preference

BenchmarkScorePositionSettingSourceDate
LMArena Text1373#142 of 297, top 48%highLMArena2026-10-08
LMArena Creative Writing1325#150 of 295, top 51%highLMArena2026-10-08
Short-Story Creative Writing83.1%#8 of 39, top 21%mediumEpoch AI
EQ-Bench Creative Writing1313#77 of 115, top 67%EQ-Bench
WildBench85.5%#8 of 57, top 15%HELM Capabilities
LMArena Multi-Turn1363#147 of 295, top 50%highLMArena2026-10-08

API pricing by provider

RouteInput $/MOutput $/MCached input $/MChecked
azure$0.25$2$0.032026-10-10
openai$0.25$2$0.0252026-10-10
openrouter$0.25$2$0.0252026-10-10

All OpenAI API prices → Estimate your cost →

Compare GPT-5 Mini

Other OpenAI models

Frequently asked questions

How good is GPT-5 Mini?

GPT-5 Mini by OpenAI ranks 128th of 354 ranked models on the Noometry Index as of October 2026, with a score of 41.8. Its strongest category is instruction following, where it ranks 46th. API pricing starts at $0.25 per million input tokens and $2 per million output tokens, with a 400K-token context window.

How much does GPT-5 Mini cost?

GPT-5 Mini costs $0.25 per million input tokens and $2 per million output tokens on OpenAI's own API, with cached input at $0.025.

What is GPT-5 Mini's context window?

GPT-5 Mini accepts up to 400K tokens of input and can write up to 128K tokens in one response.

Is GPT-5 Mini open source?

No. GPT-5 Mini is proprietary and available only through OpenAI's API and partner platforms.

How fast is GPT-5 Mini?

GPT-5 Mini generated about 3 output tokens per second in the Kagi LLM Benchmark's timed runs. Speed varies by provider, load and reasoning effort.

What are GPT-5 Mini's strengths and weaknesses?

Relative to other ranked models, GPT-5 Mini places best in instruction following, math, knowledge and lowest in multimodal, reasoning, writing & preference.

What is GPT-5 Mini best at?

Its best category is instruction following, where it ranks 46th on Noometry.