DeepSeek-V4-Pro vs MiniMax M3
Wins 3 of 7 areas
Coding · Reasoning · Math
Wins 1 of 7 areas
Long documents
DeepSeek-V4-Pro wins more areas, narrowly.MiniMax M3 is better at long documents.
Scores updated · 40 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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.
- MathCompetition and research-level math20DeepSeek-V4-Pro2 of 2 tests
- CodingWriting and fixing software43DeepSeek-V4-Pro4 of 8 tests · 1 tie
- ReasoningHard problems that need careful thinking32DeepSeek-V4-Pro3 of 5 tests
- Long documentsFinding answers in very long texts01MiniMax M31 of 1 test
- AgentsCarrying out multi-step tasks on its own22Even2 each · 2 ties
- FactsGetting facts right instead of making them up11Even1 each
- Following instructionsDoing exactly what it is asked11Even1 each
Long documents rests on a single test.
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-V4-Pro costs 13% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where DeepSeek-V4-Pro pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+26.3points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+14.8points ahead
- Fresh competition math problemsLiveBench · Mathematics+13.7points ahead
Where MiniMax M3 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+75.7points ahead
- Reasons across sets of long documentsAA-LCR+8.3points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+6.4points ahead
Every test, side by side
All 40 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V4-Pro
- Terminal-Bench HardDeepSeek-V4-Pro by 3.846.242.4+3.8
- SciCodeDeepSeek-V4-Pro by 3.750.847.1+3.7
- SWE-bench ProMiniMax M3 by 3.655.459+3.6
- LiveBench · Agentic CodingDeepSeek-V4-Pro by 1.942.640.7+1.9
- LMArena · WebDevMiniMax M3 by 19 rating points14631482+19 rating
- LiveBench · CodingDeepSeek-V4-Pro by 1.87068.2+1.8
- Terminal-Bench 2.1MiniMax M3 by 1.26465.2+1.2
- SWE-bench Verifiedtie80.680.5tie
AgentsEven
- τ-Bench V3 · BankingDeepSeek-V4-Pro by 14.830.115.3+14.8
- Terminal-Bench 4.0DeepSeek-V4-Pro by 12.614.62+12.6
- GDPValMiniMax M3 by 4.432.937.3+4.4
- AA ApexAgentsMiniMax M3 by 3.424.327.7+3.4
- MCP Atlastie73.674.2tie
- BrowseComptie83.483.5tie
ReasoningDeepSeek-V4-Pro
- CritPtDeepSeek-V4-Pro by 9.212.93.7+9.2
- LiveBench · ReasoningDeepSeek-V4-Pro by 8.282.774.5+8.2
- SimpleBenchDeepSeek-V4-Pro by 5.150.945.8+5.1
- GPQA DiamondMiniMax M3 by 4.188.892.9+4.1
- Humanity's Last ExamMiniMax M3 by 1.537.539+1.5
FactsEven
- AA-Omniscience · Non-hallucinationMiniMax M3 by 75.75.981.6+75.7
- AA-Omniscience · AccuracyDeepSeek-V4-Pro by 26.34316.7+26.3
MathDeepSeek-V4-Pro
- LiveBench · MathematicsDeepSeek-V4-Pro by 13.790.777+13.7
- HMMT Feb. 2026DeepSeek-V4-Pro by 10.895.284.4+10.8
Following instructionsEven
- IFBenchMiniMax M3 by 6.476.582.9+6.4
- LiveBench · Instruction FollowingDeepSeek-V4-Pro by 4.962.457.5+4.9
Other results14 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.
- τ³-Bench BankingDeepSeek-V4-Pro by 132613+13
- AA-OmniscienceMiniMax M3 by 12.1-10.71.4+12.1
- HMMT Nov. 2025DeepSeek-V4-Pro by 1094.484.4+10
- GDPval-AA v2MiniMax M3 by 85 rating points13061391+85 rating
- τ²-Bench Telecom (AA run)DeepSeek-V4-Pro by 7.396.288.9+7.3
- DeepSWEMiniMax M3 by 7.212.820+7.2
- LiveBenchDeepSeek-V4-Pro by 3.673.670+3.6
- NL2RepoMiniMax M3 by 3.638.542.1+3.6
- AA Agentic IndexMiniMax M3 by 3.127.730.8+3.1
- livebench_data_analysisMiniMax M3 by 1.774.576.2+1.7
- livebench_languageDeepSeek-V4-Pro by 1.378.176.8+1.3
- AA IntelligenceDeepSeek-V4-Pro by 1.230.429.2+1.2
- Artificial Analysis Coding Indextie59.458.6tie
- Tool-Decathlontie52.852.8tie
Questions people ask
Which is better, DeepSeek-V4-Pro or MiniMax M3?
DeepSeek-V4-Pro wins three of the seven areas where both have results: coding, reasoning and math. MiniMax M3 wins long documents. They are level on agents, facts and following instructions.
Which is better for coding?
DeepSeek-V4-Pro. It wins 4 of the 8 coding tests both models report; MiniMax M3 wins 3, and 1 is a tie.
Which is cheaper?
DeepSeek-V4-Pro costs $0.43 per million input tokens and $0.87 per million output tokens; MiniMax M3 costs $0.30 and $1.20. That makes DeepSeek-V4-Pro about 13% cheaper for the same work.
How do you compare the two?
We use the 40 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 14 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.