Home/Models/GPT-5 Mini
OpenAI logo

GPT-5 Mini

by OpenAI · Released Aug 2025

Multimodal
62.8
avg score
Rank #57
Compare
Better than 76% of all models
Context
400K tokens (~200 books)
Input $/1M
$0.25
Output $/1M
$2.00
Type
multimodal
License
Proprietary
Benchmarks
28 tested
Data updated today
About

GPT-5 Mini is a compact version of GPT-5, designed to handle lighter-weight reasoning tasks. It provides the same instruction-following and safety-tuning benefits as GPT-5, but with reduced latency and cost....

Tested on 28 benchmarks with 56.0% average. Top scores: MATH level 5 (97.8%), HELM — IFEval (92.7%), OTIS Mock AIME 2024-2025 (86.7%).

Looking for similar performance at lower cost?
Qwen3 30B A3B Thinking 2507 scores 63.5 (101% as good) at $0.08/1M input · 68% cheaper
Capabilities
coding
53.8
#56 globally
reasoning
50.5
#47 globally
math
60.8
#48 globally
knowledge
50.4
#99 globally
language
75.4
#57 globally
Benchmark Scores
Compare All
Tested on 28 benchmarks · Ranked across 5 categories
Score Distribution (all 233 models)
0255075100
▲ You are here
LiveBench — Coding

Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.

76.1
SWE-Bench verified

Real-world software engineering tasks from GitHub issues. Models must diagnose bugs and write patches that pass test suites. Human-verified subset of SWE-bench.

64.7
SWE-Bench Verified (Bash Only)

SWE-bench Verified solved using only bash commands, no specialized frameworks. Tests raw terminal-based problem solving.

59.8
HELM — WildBench

Stanford HELM WildBench evaluation. Tests reasoning on challenging real-world tasks.

85.5
LiveBench — Reasoning

Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.

58.6
ARC-AGI

Abstraction and Reasoning Corpus. Tests fluid intelligence through novel visual pattern recognition puzzles. Core measure of general intelligence.

54.3
MATH level 5

Competition-level math from AMC, AIME, and olympiad problems. Level 5 is the hardest tier, requiring creative problem-solving.

97.8
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

86.7
LiveBench — Mathematics

Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.

74.4
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Links
Documentation
Community
BenchGecko API
gpt-5-mini
Specifications
  • Typemultimodal
  • Context400K tokens (~200 books)
  • ReleasedAug 2025
  • LicenseProprietary
  • StatusActive
  • Cost / Message~$0.003
Available On
OpenAI logoOpenAI$0.25
Share & Export
Tweet
GPT-5 Mini is a proprietary multimodal AI model by OpenAI, released in August 2025. It has an average benchmark score of 62.8. Context window: 400K tokens.