Mercury 2 vs Seed 1.6

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

Mercury 2 and Seed 1.6 finish on the same overall score, so this is less about a single winner and more about where the edge shows up. The headline says tie; the benchmark table is where the real choice happens.

Seed 1.6 is also the more expensive model on tokens at $0.25 input / $2.00 output per 1M tokens, versus $0.25 input / $0.75 output per 1M tokens for Mercury 2. That is roughly 2.7x on output cost alone. Seed 1.6 gives you the larger context window at 256K, compared with 128K for Mercury 2.

Quick Verdict

Treat this as a split decision. Mercury 2 makes more sense if reasoning is the priority or you want the cheaper token bill; Seed 1.6 is the better fit if multimodal & grounded is the priority or you need the larger 256K context window.

Agentic

Mercury 2

Mercury 2

63.7

Seed 1.6

62.3

63
Terminal-Bench 2.0
63
67
BrowseComp
67
62
OSWorld-Verified
58

Coding

Seed 1.6

Mercury 2

41.1

Seed 1.6

42.4

75
HumanEval
64
46
SWE-bench Verified
46
38
LiveCodeBench
38
43
SWE-bench Pro
46

Multimodal & Grounded

Seed 1.6

Mercury 2

68.3

Seed 1.6

79.6

66
MMMU-Pro
80
71
OfficeQA Pro
79

Reasoning

Mercury 2

Mercury 2

80.1

Seed 1.6

74.5

82
SimpleQA
69
82
MuSR
69
87
BBH
86
77
LongBench v2
77
76
MRCRv2
78

Knowledge

Mercury 2

Mercury 2

57.2

Seed 1.6

56.4

78
MMLU
73
78
GPQA
72
76
SuperGPQA
70
74
OpenBookQA
68
72
MMLU-Pro
75
9
HLE
11
69
FrontierScience
68

Instruction Following

Seed 1.6

Mercury 2

84

Seed 1.6

87

84
IFEval
87

Multilingual

Seed 1.6

Mercury 2

79.7

Seed 1.6

83.4

81
MGSM
88
79
MMLU-ProX
81

Mathematics

Mercury 2

Mercury 2

80.9

Seed 1.6

75.9

81
AIME 2023
72
83
AIME 2024
74
82
AIME 2025
73
77
HMMT Feb 2023
68
79
HMMT Feb 2024
70
78
HMMT Feb 2025
69
80
BRUMO 2025
71
82
MATH-500
82

Frequently Asked Questions

Which is better, Mercury 2 or Seed 1.6?

Mercury 2 and Seed 1.6 are tied on overall score, so the right pick depends on which category matters most for your use case.

Which is better for knowledge tasks, Mercury 2 or Seed 1.6?

Mercury 2 has the edge for knowledge tasks in this comparison, averaging 57.2 versus 56.4. Inside this category, GPQA is the benchmark that creates the most daylight between them.

Which is better for coding, Mercury 2 or Seed 1.6?

Seed 1.6 has the edge for coding in this comparison, averaging 42.4 versus 41.1. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, Mercury 2 or Seed 1.6?

Mercury 2 has the edge for math in this comparison, averaging 80.9 versus 75.9. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, Mercury 2 or Seed 1.6?

Mercury 2 has the edge for reasoning in this comparison, averaging 80.1 versus 74.5. Inside this category, SimpleQA is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Mercury 2 or Seed 1.6?

Mercury 2 has the edge for agentic tasks in this comparison, averaging 63.7 versus 62.3. Inside this category, OSWorld-Verified is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Mercury 2 or Seed 1.6?

Seed 1.6 has the edge for multimodal and grounded tasks in this comparison, averaging 79.6 versus 68.3. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, Mercury 2 or Seed 1.6?

Seed 1.6 has the edge for instruction following in this comparison, averaging 87 versus 84. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, Mercury 2 or Seed 1.6?

Seed 1.6 has the edge for multilingual tasks in this comparison, averaging 83.4 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.