Mercury 2 vs DeepSeek-R1

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

Mercury 2 is clearly ahead on the aggregate, 65 to 43. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Mercury 2's sharpest advantage is in reasoning, where it averages 80.1 against 50.9. The single biggest benchmark swing on the page is MuSR, 82 to 40.

DeepSeek-R1 is also the more expensive model on tokens at $0.55 input / $2.19 output per 1M tokens, versus $0.25 input / $0.75 output per 1M tokens for Mercury 2. That is roughly 2.9x on output cost alone.

Quick Verdict

Pick Mercury 2 if you want the stronger benchmark profile. DeepSeek-R1 only becomes the better choice if its workflow or ecosystem matters more than the raw scoreboard.

Agentic

Mercury 2

Mercury 2

63.7

DeepSeek-R1

44.5

63
Terminal-Bench 2.0
42
67
BrowseComp
49
62
OSWorld-Verified
44

Coding

Mercury 2

Mercury 2

41.1

DeepSeek-R1

21.5

75
HumanEval
36
46
SWE-bench Verified
17
38
LiveCodeBench
19
43
SWE-bench Pro
25

Multimodal & Grounded

Mercury 2

Mercury 2

68.3

DeepSeek-R1

47.5

66
MMMU-Pro
43
71
OfficeQA Pro
53

Reasoning

Mercury 2

Mercury 2

80.1

DeepSeek-R1

50.9

82
SimpleQA
42
82
MuSR
40
87
BBH
66
77
LongBench v2
58
76
MRCRv2
57

Knowledge

Mercury 2

Mercury 2

57.2

DeepSeek-R1

37.9

78
MMLU
44
78
GPQA
43
76
SuperGPQA
41
74
OpenBookQA
39
72
MMLU-Pro
52
9
HLE
14
69
FrontierScience
44

Instruction Following

Mercury 2

Mercury 2

84

DeepSeek-R1

69

84
IFEval
69

Multilingual

Mercury 2

Mercury 2

79.7

DeepSeek-R1

60.4

81
MGSM
61
79
MMLU-ProX
60

Mathematics

Mercury 2

Mercury 2

80.9

DeepSeek-R1

52.5

81
AIME 2023
44
83
AIME 2024
46
82
AIME 2025
45
77
HMMT Feb 2023
40
79
HMMT Feb 2024
42
78
HMMT Feb 2025
41
80
BRUMO 2025
43
82
MATH-500
64

Frequently Asked Questions

Which is better, Mercury 2 or DeepSeek-R1?

Mercury 2 is ahead overall, 65 to 43. The biggest single separator in this matchup is MuSR, where the scores are 82 and 40.

Which is better for knowledge tasks, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for knowledge tasks in this comparison, averaging 57.2 versus 37.9. Inside this category, GPQA is the benchmark that creates the most daylight between them.

Which is better for coding, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for coding in this comparison, averaging 41.1 versus 21.5. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for math in this comparison, averaging 80.9 versus 52.5. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for reasoning in this comparison, averaging 80.1 versus 50.9. Inside this category, MuSR is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for agentic tasks in this comparison, averaging 63.7 versus 44.5. 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, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for multimodal and grounded tasks in this comparison, averaging 68.3 versus 47.5. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for instruction following in this comparison, averaging 84 versus 69. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, Mercury 2 or DeepSeek-R1?

Mercury 2 has the edge for multilingual tasks in this comparison, averaging 79.7 versus 60.4. 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.