GPT-5

  • Closed weights
  • Reasoning
  • Input: text · image · file
  • Released 2025-08
  • Retires 11 Dec 2026
RankedEnough results for an index.
IndexAll categories, out of 100.53.5±3.4Median 50.7
CoverageIndex weight that has results.95%
SpeedOutput tokens per second.77/sMedian 76/s
First tokenSeconds to the first output token.4.3 sMedian 1.1 s
Input / 1MUS dollars per 1M input tokens.$1.25batch $0.625Median $0.60
Output / 1MUS dollars per 1M output tokens.$10.00batch $5.00 Batch price per 1M output tokens.Median $2.50
ContextMaximum tokens in one request.400KMedian 262K
EloLMArena rating and rank.1406±4 #154Median 1425

Details

From OpenRouter
API IDThe model ID on OpenRouter.
openai/gpt-5
Max outputMaximum output tokens per request.
128K
Knowledge cutoffLast date of training data.
30 Sep 2024
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.125
Weights
—
RetiresDate OpenRouter removes the model.
11 Dec 2026

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.5 counted49.8129 / 221
CodingCode writing and repair.5 counted51.3131 / 214
ReasoningLogic problems and puzzles.7 counted54.390 / 225
MultimodalTasks with images and text.1 counted55.9—
KnowledgeFacts and expert knowledge.5 counted57.479 / 266
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted56.5—
MathMath problems.8 counted54.868 / 171

Results

32 counted
BenchmarkResultThe score from the source.PlacePlace among results on this benchmark.
MATH level 5Math98.1% ±0.31 / 88
MATH 500Math96.0% ±0.93 / 59
AIMEMath93.4% ±1.314 / 95
LiveCodeBenchCoding85.9% ±1.025 / 142
Artificial Analysis IFBenchInstruction73.1%31 / 140
FrontierMath-2025-02-28-PrivateMath32.4% ±2.816 / 69
τ²-bench RetailAgentic81.6%6 / 23
Terminal-Bench 1.0Agentic48.8% ±5.613 / 47
τ²-bench TelecomAgentic95.8%5 / 18
OTIS Mock AIME 2024-2025Math91.4% ±3.759 / 203
MMLU ProKnowledge86.5% ±0.341 / 135
GPQA diamond · Epoch AIKnowledge86.2% ±2.172 / 220
GPQA Diamond · Vals AIKnowledge85.6% ±1.845 / 135
SimpleQA VerifiedKnowledge50.1% ±1.633 / 94
Critical Physics TasksReasoning5.7%77 / 207
Artificial Analysis GPQA DiamondKnowledge85.4%73 / 193
Artificial Analysis MMLU-ProKnowledge87.1%5 / 13
IOI v1Coding20.0% ±4.624 / 61
MMMU ProMultimodal81.5% ±0.939 / 92
Chess PuzzlesReasoning36.0% ±4.845 / 93
Mystery Game PuzzlesReasoning23.0% ±4.242 / 83
ARC-AGI-1 (semi-private)Reasoning65.7%67 / 132
Artificial Analysis MMMU-ProMultimodal74.2%57 / 109
ARC-AGI-2 (semi-private)Reasoning9.9%73 / 137
τ²-bench AirlineAgentic62.5%12 / 22
Terminal-Bench 2.0 · Vals AIAgentic37.1% ±5.138 / 66
FrontierMath-Tiers-1-3-v2-PrivateMath55.4% ±2.954 / 93
SWE-Bench verified · Epoch AICoding73.6% ±2.019 / 32
GDPval-AA normalizedAgentic20.9%81 / 132
FrontierMath-Tier-4-v2-PrivateMath22.0% ±6.551 / 75
Vibe Code Bench v1.1Coding20.1% ±3.498 / 130
SWE-benchCoding69.0% ±2.167 / 87
EBR-benchReasoning12.7%26 / 33
JobBenchAgentic8.5%37 / 37
MirrorCodeCoding4.4% ±3.440 / 40
MedQAHealthcare96.3% ±0.24 / 95
CaseLaw v2Legal66.5% ±0.76 / 53
LegalBenchLegal86.0% ±0.417 / 149
FrontierMath-2025-02-28-PublicMath60.0% ±16.311 / 64
MGSMMultilingual92.8% ±1.614 / 74
MedCodeHealthcare49.6% ±2.121 / 108
TaxEval v2Finance73.4% ±0.949 / 143
MedScribeHealthcare83.7% ±1.944 / 111
MortgageTaxFinance65.5% ±0.941 / 97
CorpFin v2Finance61.1% ±1.058 / 131
Finance Agent v1.1Finance52.2% ±2.925 / 51
ARC-AGI-2 (public eval)Reasoning9.6%63 / 127
ARC-AGI-1 (public eval)Reasoning65.9%66 / 116

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseEpoch AICC BYVals AINo licence statedSierra ResearchMITARC Prize FoundationTerms restrict reuseSWE-bench teamNo licence statedmodels.devMITCheaper InferenceNo licence stated

History

13 releases
Index53.7 → 53.5
18 Nov 202621 Nov 2026
Output / 1M$10.00 → $10.00
18 Nov 202621 Nov 2026

Questions

How much does GPT-5 cost?
$1.25 per 1M input tokens, $10.00 per 1M output tokens.
How good is GPT-5?
Its index is 53.5 out of 100. That is place 114 of 282 ranked models.
What is GPT-5 best at?
Knowledge: 57.4 out of 100, place 79 of 266.
How fast is GPT-5?
77 output tokens per second, 4.3 s to the first token.
What is the context window of GPT-5?
400K tokens.
Is GPT-5 open source?
No. Its weights are not published.
When was GPT-5 released?
August 2025.

Badge

GPT-5 on Inference360: place 114 of 282, index 53.5

Compare with

GPT-5.1-Codex-Max53.6Claude Sonnet 4.5 Thinking53.3GPT-5.4 mini53.9Qwen3.5 397B53.1GPT-5 mini48.0

Lower price

dots3-note Preview62.3 · FreeApodex 1.1 Mini54.9 · FreeGLM-5.3-Flash62.4 · $0.078/1M

More from OpenAI

All models
GPT-5 mini48.0GPT-5 nano41.7GPT-OSS 120B43.2GPT-OSS 20B41.6GPT RealtimeUnranked