Model

GPT-5

CLOSED

OpenAI · GPT-5 family · released Aug 7, 2025

Unified model merging o-series reasoning with fast GPT responses via an auto-router.

ReasoningCodingVisionFunction callingTool useAgentic

2082.7

Elo · rank #54

Parameters

Undisclosed

Active params

—

Context

400K tokens

Architecture

Unified reasoning/non-reasoning system behind a real-time auto-router; dense transformer, exact size undisclosed

License

Proprietary (OpenAI API Terms)

Languages

—

API price (in/out)

$1.25 / $10

Modalities

text · vision

Benchmark results

Bar shows position within the tracked field; marker = field best

Aider PolyglotCoding88.0%#2

best: Claude Opus 4.5 · 89.4%

AIMEMath94.6%#20

best: GPT-5.2 · 100.0%

ARC-AGI-1Reasoning65.7%#11

best: GPT-5.6 · 96.5%

ARC-AGI-2Reasoning9.9%#24

best: GPT-5.6 · 92.5%

FrontierMath (Tiers 1-3)Math24.8%#10

best: GPT-5.6 · 89.0%

FrontierMath (Tier 4)Math8.3%#8

best: GPT-5.6 · 83.0%

GAIAAgents59.4%#4

best: MiniMax-M2 · 75.7%

GPQA DiamondReasoning85.7%#55

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning24.8%#57

best: Claude Sonnet 5 · 57.4%

LiveCodeBenchCoding67.6%#74

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MATH-500Math99.4%#1

best: this model · 99.4%

MMMU-ProVision78.4%#15

best: Claude Opus 4.7 · 85.5%

MMMUVision84.2%#10

best: Claude Fable 5 · 89.3%

SimpleQAKnowledge55.0%#2

best: GPT-4.5 · 62.5%

SWE-bench VerifiedCoding74.9%#42

best: Claude Fable 5 · 95.0%

τ²-Bench TelecomAgents96.7%#13

best: Claude Opus 4.6 · 99.3%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$1.25

Output / M tok

$10

GPT-5 family

Elo progression across releases

GPT-5Aug 20252082.7 GPT-5 miniAug 20251873.5 GPT-5 nanoAug 20251595.3 GPT-5 ProAug 20252349.9 GPT-5-CodexSep 20252103.4

Resources

DOCSdevelopers.openai.com/api/docs/models/gpt-5 APIAvailable via OpenAI API

API price $1.25/$10 · each benchmark row carries its own source badge (see methodology)