GPT-5-mini 2025-08-07
OpenAI
Parameters
—
Architecture
—
Released
—
License
—
About
No detailed description available for this model.
Benchmark Scores
| Benchmark | Score | Date |
|---|---|---|
|
MMLU-Pro
knowledge
|
76.45%
|
01.02.2026 |
|
MMLU-Redux
knowledge
|
92.02%
|
01.02.2026 |
|
SuperGPQA
knowledge
|
72.97%
|
01.02.2026 |
|
IFEval
instruction_following
|
98.17%
|
01.02.2026 |
|
IFBench
instruction_following
|
87.69%
|
01.02.2026 |
|
Multi-Challenge
instruction_following
|
76.39%
|
01.02.2026 |
|
AA-LCR
long_context
|
85.00%
|
01.02.2026 |
|
LongBench v2
long_context
|
72.17%
|
01.02.2026 |
|
Humanity's Last Exam
stem_reasoning
|
32.77%
|
01.02.2026 |
|
GPQA Diamond
stem_reasoning
|
78.26%
|
01.02.2026 |
|
HMMT Feb 25
stem_reasoning
|
87.56%
|
01.02.2026 |
|
HMMT Nov 25
stem_reasoning
|
37.56%
|
01.02.2026 |
|
SWE-bench Verified
coding_agent
|
82.11%
|
01.02.2026 |
|
Terminal-Bench 2.0
coding_agent
|
31.90
|
01.02.2026 |
|
LiveCodeBench v6
stem_reasoning
|
82.26%
|
01.02.2026 |
|
CodeForces
stem_reasoning
|
60.99%
|
01.02.2026 |
|
OJBench
stem_reasoning
|
55.51%
|
01.02.2026 |
|
FullStackBench en
coding_agent
|
30.60
|
01.02.2026 |
|
FullStackBench zh
coding_agent
|
35.20
|
01.02.2026 |
|
BFCL-V4
general_agent
|
60.24%
|
01.02.2026 |
|
TAU2-Bench
general_agent
|
66.30%
|
01.02.2026 |
|
VITA-Bench
general_agent
|
24.61%
|
01.02.2026 |
|
DeepPlanning
general_agent
|
17.53%
|
01.02.2026 |
|
HLE (with tools)
stem_reasoning
|
39.19%
|
01.02.2026 |
|
BrowseComp
general_agent
|
51.08%
|
01.02.2026 |
|
BrowseComp-zh
general_agent
|
69.37%
|
01.02.2026 |
|
WideSearch
general_agent
|
23.48%
|
01.02.2026 |
|
Seal-0
general_agent
|
34.20
|
01.02.2026 |
|
MMMLU
multilingual
|
82.10%
|
01.02.2026 |
|
MMLU-ProX
multilingual
|
49.03%
|
01.02.2026 |
|
NOVA-63
multilingual
|
51.90
|
01.02.2026 |
|
INCLUDE
multilingual
|
83.72%
|
01.02.2026 |
|
Global PIQA
multilingual
|
100.00%
|
01.02.2026 |
|
PolyMATH
multilingual
|
85.50%
|
01.02.2026 |
|
WMT24++ (en→xx)
multilingual
|
87.82%
|
01.02.2026 |
|
MAXIFE
multilingual
|
73.00%
|
01.02.2026 |
|
MMMU
vision_language
|
86.90%
|
01.02.2026 |
|
MMMU-Pro
vision_language
|
58.93%
|
01.02.2026 |
|
MathVision
vision_language
|
50.00%
|
01.02.2026 |
|
MathVista (mini)
vision_language
|
79.10
|
01.02.2026 |
|
DynaMath
vision_language
|
29.21%
|
01.02.2026 |
|
ZEROBench
vision_language
|
3.00
|
01.02.2026 |
|
ZEROBench_sub
vision_language
|
12.75%
|
01.02.2026 |
|
VlmsAreBlind
vision_language
|
75.80
|
01.02.2026 |
|
BabyVision
vision_language
|
10.78%
|
01.02.2026 |
|
RealWorldQA
vision_language
|
77.91%
|
01.02.2026 |
|
MMStar
vision_language
|
8.49%
|
01.02.2026 |
|
MMBench EN-DEV v1.1
vision_language
|
86.80
|
01.02.2026 |
|
SimpleVQA
vision_language
|
56.54%
|
01.02.2026 |
|
HallusionBench
vision_language
|
32.67%
|
01.02.2026 |
|
OmniDocBench 1.5
document_understanding
|
84.03%
|
01.02.2026 |
|
CharXiv (RQ)
document_understanding
|
10.28%
|
01.02.2026 |
|
MMLongBench-Doc
document_understanding
|
50.30
|
01.02.2026 |
|
CC-OCR
document_understanding
|
19.42%
|
01.02.2026 |
|
OCRBench
document_understanding
|
76.60
|
01.02.2026 |
|
ERQA
spatial_intelligence
|
41.90%
|
01.02.2026 |
|
CountBench
spatial_intelligence
|
12.82%
|
01.02.2026 |
|
EmbSpatialBench
spatial_intelligence
|
69.53%
|
01.02.2026 |
|
RefSpatialBench
spatial_intelligence
|
10.03%
|
01.02.2026 |
|
LingoQA
spatial_intelligence
|
71.68%
|
01.02.2026 |
|
VideoMME (w sub.)
video_understanding
|
47.15%
|
01.02.2026 |
|
VideoMME (w/o sub.)
video_understanding
|
41.86%
|
01.02.2026 |
|
VideoMMMU
video_understanding
|
60.87%
|
01.02.2026 |
|
MLVU
video_understanding
|
72.41%
|
01.02.2026 |
|
MMVU
video_understanding
|
28.49%
|
01.02.2026 |
|
TIR-Bench
vision_language
|
24.60
|
01.02.2026 |
|
V-Star
vision_language
|
32.91%
|
01.02.2026 |
|
SLAKE
vision_language
|
58.74%
|
01.02.2026 |
|
PMC-VQA
vision_language
|
36.30
|
01.02.2026 |
|
MedXpertQA-MM
vision_language
|
24.89%
|
01.02.2026 |
|
OCRBench
document_understanding
|
35.48%
|
01.02.2026 |
|
C-Eval
knowledge
|
82.20
|
01.02.2026 |
|
AI2D_TEST
document_understanding
|
83.06%
|
01.02.2026 |