GPT-5 mini vs Mercury 2

Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.

GPT-5 mini is clearly ahead on the aggregate, 69 to 65. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

GPT-5 mini's sharpest advantage is in multimodal & grounded, where it averages 83.8 against 68.3. The single biggest benchmark swing on the page is MMMU-Pro, 86 to 66. Mercury 2 does hit back in instruction following, so the answer changes if that is the part of the workload you care about most.

Quick Verdict

Pick GPT-5 mini if you want the stronger benchmark profile. Mercury 2 only becomes the better choice if instruction following is the priority.

Agentic

GPT-5 mini

GPT-5 mini

65.7

Mercury 2

63.7

68
Terminal-Bench 2.0
63
70
BrowseComp
67
60
OSWorld-Verified
62

Coding

GPT-5 mini

GPT-5 mini

42.8

Mercury 2

41.1

80
HumanEval
75
41
SWE-bench Verified
46
37
LiveCodeBench
38
49
SWE-bench Pro
43

Multimodal & Grounded

GPT-5 mini

GPT-5 mini

83.8

Mercury 2

68.3

86
MMMU-Pro
66
81
OfficeQA Pro
71

Reasoning

GPT-5 mini

GPT-5 mini

81.8

Mercury 2

80.1

84
SimpleQA
82
82
MuSR
82
87
BBH
87
80
LongBench v2
77
79
MRCRv2
76

Knowledge

GPT-5 mini

GPT-5 mini

62.8

Mercury 2

57.2

88
MMLU
78
86
GPQA
78
84
SuperGPQA
76
82
OpenBookQA
74
73
MMLU-Pro
72
16
HLE
9
75
FrontierScience
69

Instruction Following

Mercury 2

GPT-5 mini

82

Mercury 2

84

82
IFEval
84

Multilingual

GPT-5 mini

GPT-5 mini

80.1

Mercury 2

79.7

82
MGSM
81
79
MMLU-ProX
79

Mathematics

GPT-5 mini

GPT-5 mini

87.2

Mercury 2

80.9

90
AIME 2023
81
92
AIME 2024
83
91
AIME 2025
82
86
HMMT Feb 2023
77
88
HMMT Feb 2024
79
87
HMMT Feb 2025
78
89
BRUMO 2025
80
85
MATH-500
82

Frequently Asked Questions

Which is better, GPT-5 mini or Mercury 2?

GPT-5 mini is ahead overall, 69 to 65. The biggest single separator in this matchup is MMMU-Pro, where the scores are 86 and 66.

Which is better for knowledge tasks, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for knowledge tasks in this comparison, averaging 62.8 versus 57.2. Inside this category, MMLU is the benchmark that creates the most daylight between them.

Which is better for coding, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for coding in this comparison, averaging 42.8 versus 41.1. Inside this category, SWE-bench Pro is the benchmark that creates the most daylight between them.

Which is better for math, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for math in this comparison, averaging 87.2 versus 80.9. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for reasoning in this comparison, averaging 81.8 versus 80.1. Inside this category, LongBench v2 is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for agentic tasks in this comparison, averaging 65.7 versus 63.7. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for multimodal and grounded tasks in this comparison, averaging 83.8 versus 68.3. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, GPT-5 mini or Mercury 2?

Mercury 2 has the edge for instruction following in this comparison, averaging 84 versus 82. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, GPT-5 mini or Mercury 2?

GPT-5 mini has the edge for multilingual tasks in this comparison, averaging 80.1 versus 79.7. Inside this category, MGSM is the benchmark that creates the most daylight between them.

Last updated: March 12, 2026

Weekly LLM Benchmark Digest

Get notified when new models drop, benchmark scores change, or the leaderboard shifts. One email per week.

Free. No spam. Unsubscribe anytime. We only store derived location metadata for consent routing.