DeepSeek-V3.2 vs Mercury 2
Wins 5 of 6 areas
Coding · Agents · Reasoning · Facts · Long documents
Wins 1 of 6 areas
Following instructions
DeepSeek-V3.2 is the stronger all-rounder.Mercury 2 is better at following instructions.
Scores updated · 24 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software40DeepSeek-V3.24 of 4 tests
- ReasoningHard problems that need careful thinking30DeepSeek-V3.23 of 3 tests
- FactsGetting facts right instead of making them up30DeepSeek-V3.23 of 3 tests
- AgentsCarrying out multi-step tasks on its own20DeepSeek-V3.22 of 2 tests
- Long documentsFinding answers in very long texts10DeepSeek-V3.21 of 1 test
- Following instructionsDoing exactly what it is asked01Mercury 21 of 1 test
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
DeepSeek-V3.2 costs 30% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where DeepSeek-V3.2 pulls ahead
- Reasons across sets of long documentsAA-LCR+29.6points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+19.5points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+11.8points ahead
Where Mercury 2 pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+9.1points ahead
Every test, side by side
All 24 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V3.2
- LMArena · WebDevDeepSeek-V3.2 by 196 rating points13621166+196 rating
- Terminal-Bench 2.1DeepSeek-V3.2 by 19.546.827.3+19.5
- Terminal-Bench HardDeepSeek-V3.2 by 9.135.626.5+9.1
- SciCodeDeepSeek-V3.2 by 1.238.937.7+1.2
AgentsDeepSeek-V3.2
- GDPValDeepSeek-V3.2 by 9.89.80+9.8
- τ-Bench V3 · BankingDeepSeek-V3.2 by 9.318.89.5+9.3
ReasoningDeepSeek-V3.2
- Humanity's Last ExamDeepSeek-V3.2 by 7.524.617.1+7.5
- GPQA DiamondDeepSeek-V3.2 by 78477+7
- CritPtDeepSeek-V3.2 by 2.12.90.8+2.1
FactsDeepSeek-V3.2
- AA-Omniscience · AccuracyDeepSeek-V3.2 by 11.83321.2+11.8
- AA-Omniscience · Non-hallucinationDeepSeek-V3.2 by 8.517.38.8+8.5
- Vectara HHEM hallucination ratelower is betterDeepSeek-V3.2 by 66.312.3+6
Following instructionsMercury 2
- IFBenchMercury 2 by 9.160.769.8+9.1
Other results10 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA-OmniscienceDeepSeek-V3.2 by 28.2-22.5-50.7+28.2
- τ²-Bench Telecom (AA run)DeepSeek-V3.2 by 19.890.670.8+19.8
- LiveCodeBenchDeepSeek-V3.2 by 16.383.367+16.3
- AA Agentic IndexDeepSeek-V3.2 by 14.318.34+14.3
- Artificial Analysis Coding IndexDeepSeek-V3.2 by 13.144.231.1+13.1
- vectara_avg_summary_lengthMercury 2 by 87.1 rating points62149.1+87.1 rating
- AA IntelligenceDeepSeek-V3.2 by 7.721.513.8+7.7
- vectara_answer_rateMercury 2 by 7.492.6100+7.4
- vectara_factual_consistencyDeepSeek-V3.2 by 693.787.7+6
- AIME 2025DeepSeek-V3.2 by 293.191.1+2
Questions people ask
Which is better, DeepSeek-V3.2 or Mercury 2?
DeepSeek-V3.2 wins five of the six areas where both have results: coding, agents, reasoning, facts and long documents. Mercury 2 wins following instructions.
Which is better for coding?
DeepSeek-V3.2. It wins 4 of the 4 coding tests both models report; Mercury 2 wins none.
Which is cheaper?
DeepSeek-V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; Mercury 2 costs $0.25 and $0.75. That makes DeepSeek-V3.2 about 30% cheaper for the same work.
How do you compare the two?
We use the 24 benchmark tests both models have published scores on. The verdict counts the 14 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 10 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.