Benchmark	Category	Top Score	Info	Self Reported	Source
Aider-Polyglot	code	0.72	High Reasoning Effort, Diff Method	Yes	Source
AIME 2025	math	0.91	High Reasoning Effort, No Tools	Yes	Source
ARC-AGI-1	-	0.54	High Reasoning Effort	No	Source
ARC-AGI-2	-	0.04	High Reasoning Effort	No	Source
BrowseComp Long Context 128k	-	0.89	High Reasoning Effort	Yes	Source
BrowseComp Long Context 256k	-	0.86	High Reasoning Effort	Yes	Source
CharXiv-Reasoning	-	0.76	High Reasoning Effort	Yes	Source
COLLIE	-	0.98	High Reasoning Effort	Yes	Source
Confabulations	-	13.28	-	No	Source
Creative Story Writing	-	8.31	-	No	Source
ERQA	-	0.63	High Reasoning Effort	Yes	Source
FActScore hallucination rate	hallucinations	0.04	High Reasoning Effort	Yes	Source
Frontier Math	math	0.22	With Thinking, With Python, Pass @ 1	Yes	Source
GPQA Diamond	general-knowledge	0.82	High Reasoning Effort, No Tools	Yes	Source
Graphwalks bfs <128k	-	0.73	High Reasoning Effort	Yes	Source
Graphwalks parents <128k	-	0.64	High Reasoning Effort	Yes	Source
HMMT 2025	-	0.88	High Reasoning Effort, No Tools	Yes	Source
Humanity's Last Exam	-	0.17	High Reasoning Effort, No Tools	Yes	Source
LongFact-Concepts hallucination rate	hallucinations	0.01	High Reasoning Effort	Yes	Source
LongFact-Objects hallucination rate	hallucinations	0.01	High Reasoning Effort	Yes	Source
MathArena Apex	-	0.01	High Reasoning Effort	No	Source
MMLU Pro	-	0.62	High Reasoning Effort	Yes	Source
MMMU	-	0.82	High Reasoning Effort	Yes	Source
MMMU Pro	-	0.74	High Reasoning Effort	Yes	Source
OpenAI-MRCR: 2 needle 128k	-	0.84	High Reasoning Effort	Yes	Source
OpenAI-MRCR: 2 needle 256k	-	0.59	High Reasoning Effort	Yes	Source
SWE-Bench	code	0.71	High Reasoning Effort, No Tools	Yes	Source
Tau 2 Airline	-	0.60	High Reasoning Effort	Yes	Source
Tau 2 Retail	-	0.78	High Reasoning Effort	Yes	Source
Tau 2 Telecom	-	0.74	High Reasoning Effort	Yes	Source
Video MMMU	-	0.82	High Reasoning Effort	Yes	Source
VideoMME	-	0.79	High Reasoning Effort	Yes	Source

Benchmark comparisons

Use the selector to switch benchmarks and see how this model stacks up against its closest competitors.

OpenAI-MRCR: 2 needle 128k

Compare this model with the leading peers for the selected benchmark.

Benchmark

0.84

Rank #2/8

8 models

Showing 8 models around the selected model (out of 8 total).

View benchmark page

Preparing your page

GPT 5 mini

OpenAI

Highlights

Benchmark table

Benchmark comparisons

OpenAI-MRCR: 2 needle 128k