0

Claude Opus 4.5 vs GPT-5 Mini

Head to head on 31 shared benchmarks, with price, context window, and release dates.

Benchmark scores

75

evals · sort by any column, ⤢ to expand fullscreen

Model
Claude Opus 4.5

Anthropic

62.7%61.3%35.6%6.1%1683elo73.120.7%81.0%13.2%43.0%78.1%79.7%74.4%62.5%81.3%90.4%80.1%73.8%45.2%83.2%88.9%68.7%36.0%45.0%47.0%70.7%79.2%74.3%40.9%20.6%86.3%------16.7%---------84.7%--12.13-83.8%-83.5%----100.0%37.477.5%--45.361.8---------62.2%
GPT-5 Mini

OpenAI

46.7%60.2%18.1%2.8%1310elo56.227.2%68.7%5.1%45.6%69.1%68.2%55.2%65.3%75.5%82.2%68.3%54.5%43.0%80.6%77.5%66.9%9.0%43.0%36.9%39.7%59.8%75.2%14.4%14.2%31.9%0.0%0.0%1.130.0%-0.1520.0%-25.0%40.0%1.7100.0%100.0%66.7%0.0%88.3%60.0%-100.0%90.0%-47.2%-80.0%-0.0%1.541.5480.0%---16.8%4.95--0.0%9.5%0.0%86.1%0.4939.21272.1186.0%86.0%47.9%
2 / 2 models
AttributeClaude Opus 4.5

Anthropic

GPT-5 Mini

OpenAI

DescriptionClaude Opus 4.5 is an AI model from Anthropic.GPT-5 mini is an AI model from OpenAI.
FamilyClaudeGPT
Params--
Open weightsclosedclosed
Licenseproprietaryproprietary
Released24 Nov 20257 Aug 2025
Context200,000400,000
Input $/1M$5.00$0.25
Output $/1M$25.00$2.00
Cache read $/1M$0.500$0.025
Cache write $/1M$6.250-
Speed0 tok/s0 tok/s
Latency0.00 s0.00 s
Modalitiesfile, image, texttext, image, file
Providers--
PublisherAnthropicOpenAI
Reported scores

FAQ

Is Claude Opus 4.5 or GPT-5 Mini better?
Across the 31 benchmarks both models report on Sophon, Claude Opus 4.5 leads on 27 of them. Which one is "better" depends on the benchmark - the table above shows every shared score.
How do Claude Opus 4.5 and GPT-5 Mini compare on GPQA Diamond?
Claude Opus 4.5 scores 81.0% and GPT-5 Mini scores 68.7% on GPQA Diamond.
Which is cheaper, Claude Opus 4.5 or GPT-5 Mini?
GPT-5 Mini is cheaper at $2.00 per million output tokens against $25.00 for Claude Opus 4.5 - about 12.5x.
When were Claude Opus 4.5 and GPT-5 Mini released?
Claude Opus 4.5 was released 24 Nov 2025 by Anthropic. GPT-5 Mini was released 7 Aug 2025 by OpenAI.

Comparing something else? Build your own side-by-side.