Model

GPT-5 mini

CLOSED

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

Cost-efficient mid-tier of the GPT-5 family.

ReasoningCodingVisionFunction callingTool useAgentic

1846.1

Elo · rank #92

Parameters

Undisclosed

Active params

—

Context

400K tokens

Architecture

Cost-efficient mid-tier sibling of GPT-5's unified architecture; dense transformer, exact size undisclosed

License

Proprietary (OpenAI API Terms)

Languages

—

API price (in/out)

$0.25 / $2

Modalities

text · vision

Benchmark results

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

Aider PolyglotCoding71.6%#14

best: Claude Opus 4.5 · 89.4%

AIMEMath91.1%#44

best: GPT-5.2 · 100.0%

GDPval-AAAgents940#44

best: Claude Fable 5 · 1932

GPQA DiamondReasoning82.3%#76

best: GPT-6 Astra · 96.0%

Humanity's Last ExamReasoning16.7%#77

best: Claude Opus 5 · 64.7%

IFBenchReasoning75.0%#21

best: MiniMax M3 · 83.0%

LiveCodeBenchCoding86.6%#19

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

MATH-500Math94.8%#33

best: GPT-5 · 99.4%

MMMU-ProVision67.5%#40

best: Claude Opus 4.7 · 85.5%

OSWorld-VerifiedAgents42.0%#28

best: Claude Fable 5 · 85.0%

SWE-bench ProCoding45.7%#53

best: Claude Fable 5.1 · 81.2%

SWE-bench VerifiedCoding60.8%#80

best: Claude Opus 5 · 96.0%

τ²-Bench TelecomAgents74.1%#27

best: Claude Opus 4.6 · 99.3%

Terminal-Bench 2.0Coding38.2%#68

best: Gemini 3.8 Flash · 89.4%

Run it locally

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

Input / M tok

$0.25

Output / M tok

GPT-5 family

Elo progression across releases

GPT-5Aug 20252074.5 GPT-5 miniAug 20251846.1 GPT-5 nanoAug 20251583.9 GPT-5 ProAug 20252355.3 GPT-5-CodexSep 20252071.8

Resources

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

API price $0.25/$2 · each benchmark row carries its own source badge (see methodology)