Noometry

OpenAI, proprietary

GPT-5 Nano

GPT-5 Nano by OpenAI ranks 241st of 354 ranked models on the Noometry Index as of October 2026, with a score of 33.5. Its strongest category is instruction following, where it ranks 79th. API pricing starts at $0.05 per million input tokens and $0.40 per million output tokens, with a 400K-token context window.

Last verified

Noometry rank
#241 of 354
Index score
33.5
Evidence
Confirmed 49 results
Provider
OpenAI
Released
August 7, 2025
Weights
Proprietary
Reasoning
Yes
Context window
400K
Max output
128K
Input price
$0.05 / M
Output price
$0.40 / M
Blended price
$0.14 / M
Output speed
4 tokens/s Kagi
Value
#31 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 33.6
  2. Agentic & Tool Use 25.8
  3. Reasoning 16.3
  4. Math 29.4
  5. Knowledge 35.9
  6. Multimodal 31.3
  7. Multilingual 45.3
  8. Instruction Following 75.0
  9. Long Context 31.3
  10. Writing & Preference 39.1
CategoryScoreRankResults
Coding33.6#2543
Agentic & Tool Use25.8#1062
Reasoning16.3#3068
Math29.4#2417
Knowledge35.9#1786
Multimodal31.3#1082
Multilingual45.3#1721
Instruction Following75.0#792
Long Context31.3#2812
Writing & Preference39.1#2495

Strengths and weaknesses

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

Strongest categories

CategoryScorevs medianRank
Instruction Following75.0+3.7#79 of 305, top 26%
Knowledge35.9−1.4#178 of 314, top 57%
Multilingual45.3−2.1#172 of 297, top 58%

Weakest categories

CategoryScorevs medianRank
Long Context31.3−9.7#281 of 296, top 95%
Reasoning16.3−7.3#306 of 350, top 88%
Multimodal31.3−7.2#108 of 128, top 85%

Closest competitors

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

ModelRankScoreBlended $/MSpeed
Codellama 70b Instruct#23733.7——Compare
Qwen3 8B#23833.7$0.31—Compare
Grok-2 (Dec 2024)#23933.7——Compare
GPT-4.1 mini#24033.6$0.7086Compare
Mercury 2.5#24233.5$0.0675—Compare
Mistral Small#24333.4$0.26120Compare
Nova 2.0 Pro Preview#24433.4——Compare
Qwen2.5-Coder-32B#24533.4$0.74—Compare

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 Verified (bash only)34.8%#31 of 39, top 80%mediumSWE-bench2025-08-07
WeirdML38.1%#84 of 119, top 71%highEpoch AI
WeirdML25.9%lowEpoch AI
LMArena Coding1351#172 of 294, top 59%highLMArena2026-10-08
ALE-Bench718.67#68 of 105, top 65%highEpoch AI

Agentic & Tool Use

BenchmarkScorePositionSettingSourceDate
Terminal-Bench21.8%#37 of 41, top 91%Epoch AI
Terminal-Bench11.5%mediumEpoch AI
Berkeley Function Calling Leaderboard51.5%#18 of 49, top 37%fcBerkeley Function Calling Leaderboard

Reasoning

BenchmarkScorePositionSettingSourceDate
ARC-AGI-22.6%#64 of 83, top 78%highEpoch AI
ARC-AGI-20%lowEpoch AI
ARC-AGI-20.9%mediumEpoch AI
ARC-AGI-20%minimalEpoch AI
Kagi LLM Benchmark62.2%#39 of 99, top 40%Kagi LLM Benchmark
ARC-AGI-116.7%highEpoch AI
ARC-AGI-14%lowEpoch AI
ARC-AGI-120.7%#69 of 83, top 84%mediumEpoch AI
ARC-AGI-11.5%minimalEpoch AI
Chess Puzzles27%#39 of 129, top 31%highEpoch AI2026-08-07
Chess Puzzles15%lowEpoch AI2026-07-13
Chess Puzzles1%minimalEpoch AI2026-07-13
LMArena Hard Prompts1328#175 of 297, top 59%highLMArena2026-10-08
Mystery Game Puzzles8%highEpoch AI2026-08-27
Mystery Game Puzzles9%#62 of 74, top 84%lowEpoch AI2026-08-27
Mystery Game Puzzles9%mediumEpoch AI2026-08-27
Mystery Game Puzzles5%minimalEpoch AI2026-08-27
DTBench62.7%#109 of 151, top 73%highEpoch AI
LMCA7.9%#117 of 125, top 94%highEpoch AI
Epoch Capabilities Index139.38#112 of 213, top 53%Epoch AI2025-08-07
ForecastBench59.1#42 of 72, top 59%Epoch AI

Math

BenchmarkScorePositionSettingSourceDate
FrontierMath (Tiers 1-3)20%#69 of 81, top 86%highEpoch AI2026-06-12
FrontierMath (Tiers 1-3)6%lowEpoch AI2026-08-27
FrontierMath (Tiers 1-3)1.8%minimalEpoch AI2026-08-27
FrontierMath Tier 42.4%#60 of 63, top 96%highEpoch AI2026-06-12
OTIS Mock AIME 2024-202581.1%#79 of 173, top 46%highEpoch AI2025-10-31
OTIS Mock AIME 2024-202546.7%lowEpoch AI2026-07-13
OTIS Mock AIME 2024-202574.2%mediumEpoch AI2025-08-07
OTIS Mock AIME 2024-202535.6%minimalEpoch AI2026-07-13
ProofBench12%#61 of 77, top 80%highEpoch AI
Omni-MATH54.6%#15 of 57, top 27%HELM Capabilities
LMArena Math1317#171 of 285, top 60%highLMArena2026-10-08
MATH Level 594.9%highEpoch AI2025-08-20
MATH Level 595.2%#11 of 79, top 14%mediumEpoch AI2025-08-20
FrontierMath (Feb 2025 set)8.3%#40 of 68, top 59%highEpoch AI2025-10-30
FrontierMath (Feb 2025 set)7.2%mediumEpoch AI2025-08-07
FrontierMath Tier 4 (v1)0%highEpoch AI2025-10-30
FrontierMath Tier 4 (v1)2.1%#40 of 55, top 73%mediumEpoch AI2025-08-07

Knowledge

BenchmarkScorePositionSettingSourceDate
GPQA Diamond69.4%#99 of 186, top 54%highEpoch AI2025-10-30
GPQA Diamond57.6%lowEpoch AI2026-07-13
GPQA Diamond67.4%mediumEpoch AI2025-08-07
GPQA Diamond48.5%minimalEpoch AI2026-07-13
SimpleQA Verified11.7%#74 of 77, top 97%highEpoch AI2026-08-10
MMLU-Pro77.8%#23 of 58, top 40%HELM Capabilities
Vectara Hallucination Rate (lower is better)10.5%#58 of 96, top 61%Vectara Hallucination Leaderboard
GPQA (HELM)67.9%#13 of 57, top 23%HELM Capabilities
LMArena Expert1321#168 of 273, top 62%highLMArena2026-10-08

Multimodal

BenchmarkScorePositionSettingSourceDate
LMArena Vision1159#96 of 122, top 79%highLMArena2026-10-09
VPCT37.2%#17 of 24, top 71%highEpoch AI
VPCT35.4%mediumEpoch AI

Multilingual

BenchmarkScorePositionSettingSourceDate
LMArena Non-English1313#172 of 297, top 58%highLMArena2026-10-08
LMArena Chinese1356#165 of 285, top 58%highLMArena2026-10-08
LMArena German1327#143 of 231, top 62%highLMArena2026-10-08
LMArena Japanese1226#148 of 211, top 71%highLMArena2026-10-08
LMArena Korean1269#140 of 213, top 66%highLMArena2026-10-08
LMArena Russian1296#179 of 283, top 64%highLMArena2026-10-08
LMArena Spanish1360#135 of 226, top 60%highLMArena2026-10-08

Instruction Following

BenchmarkScorePositionSettingSourceDate
IFEval93.2%#4 of 57, top 8%HELM Capabilities
LMArena Instruction Following1306#172 of 298, top 58%highLMArena2026-10-08

Long Context

BenchmarkScorePositionSettingSourceDate
Fiction.LiveBench44.4%#40 of 47, top 86%mediumEpoch AI
LMArena Longer Query1312#179 of 291, top 62%highLMArena2026-10-08

Writing & Preference

BenchmarkScorePositionSettingSourceDate
LMArena Text1320#178 of 297, top 60%highLMArena2026-10-08
LMArena Creative Writing1249#205 of 295, top 70%highLMArena2026-10-08
EQ-Bench Creative Writing705#111 of 115, top 97%EQ-Bench
WildBench80.6%#27 of 57, top 48%HELM Capabilities
LMArena Multi-Turn1311#182 of 295, top 62%highLMArena2026-10-08

API pricing by provider

RouteInput $/MOutput $/MCached input $/MChecked
azure$0.05$0.40$0.012026-10-10
openai$0.05$0.40$0.0052026-10-10
openrouter$0.05$0.40$0.0052026-10-10

All OpenAI API prices → Estimate your cost →

Compare GPT-5 Nano

Other OpenAI models

Frequently asked questions

How good is GPT-5 Nano?

GPT-5 Nano by OpenAI ranks 241st of 354 ranked models on the Noometry Index as of October 2026, with a score of 33.5. Its strongest category is instruction following, where it ranks 79th. API pricing starts at $0.05 per million input tokens and $0.40 per million output tokens, with a 400K-token context window.

How much does GPT-5 Nano cost?

GPT-5 Nano costs $0.05 per million input tokens and $0.40 per million output tokens on OpenAI's own API, with cached input at $0.005.

What is GPT-5 Nano's context window?

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

Is GPT-5 Nano open source?

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

How fast is GPT-5 Nano?

GPT-5 Nano generated about 4 output tokens per second in the Kagi LLM Benchmark's timed runs. Speed varies by provider, load and reasoning effort.

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

Relative to other ranked models, GPT-5 Nano places best in instruction following, knowledge, multilingual and lowest in long context, reasoning, multimodal.

What is GPT-5 Nano best at?

Its best category is instruction following, where it ranks 79th on Noometry.