DeepSeek-V4-Pro vs Qwen3.7 Max
Wins 1 of 7 areas
Agents
Wins 6 of 7 areas
Coding · Reasoning · Facts · Math · Long documents · Following instructions
Qwen3.7 Max is the stronger all-rounder.DeepSeek-V4-Pro is cheaper and better at agents.
Scores updated · 48 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.
- CodingWriting and fixing software17Qwen3.7 Max7 of 10 tests · 2 ties
- ReasoningHard problems that need careful thinking03Qwen3.7 Max3 of 5 tests · 2 ties
- MathCompetition and research-level math13Qwen3.7 Max3 of 6 tests · 2 ties
- Following instructionsDoing exactly what it is asked02Qwen3.7 Max2 of 2 tests
- FactsGetting facts right instead of making them up12Qwen3.7 Max2 of 3 tests
- Long documentsFinding answers in very long texts01Qwen3.7 Max1 of 1 test
- AgentsCarrying out multi-step tasks on its own32DeepSeek-V4-Pro3 of 5 tests
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 87% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3.7 Max pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+68.5points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+31.8points ahead
- Common-sense trick questionsSimpleBench+19.5points ahead
Where DeepSeek-V4-Pro pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+18.3points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+13.1points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+11.9points ahead
Every test, side by side
All 48 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.7 Max
- Terminal-Bench 2.1Qwen3.7 Max by 10.56474.5+10.5
- LMArena · WebDevQwen3.7 Max by 52 rating points14631515+52 rating
- SWE-bench ProQwen3.7 Max by 5.255.460.6+5.2
- Terminal-Bench HardQwen3.7 Max by 4.646.250.8+4.6
- LiveBench · CodingQwen3.7 Max by 4.27074.2+4.2
- SWE-bench MultilingualQwen3.7 Max by 2.176.278.3+2.1
- SciCodeDeepSeek-V4-Pro by 1.350.849.5+1.3
- LiveBench · Agentic CodingQwen3.7 Max by 142.643.6+1
- LiveCodeBench v6tie92.591.6tie
- SWE-bench Verifiedtie80.680.4tie
AgentsDeepSeek-V4-Pro
- τ-Bench V3 · BankingDeepSeek-V4-Pro by 18.330.111.8+18.3
- Terminal-Bench 4.0DeepSeek-V4-Pro by 13.114.61.5+13.1
- AA IT-Bench SREQwen3.7 Max by 4.238.342.5+4.2
- MCP AtlasQwen3.7 Max by 2.873.676.4+2.8
- GDPValDeepSeek-V4-Pro by 1.332.931.6+1.3
ReasoningQwen3.7 Max
- SimpleBenchQwen3.7 Max by 19.550.970.4+19.5
- GPQA DiamondQwen3.7 Max by 3.588.892.3+3.5
- Humanity's Last ExamQwen3.7 Max by 337.540.5+3
- LiveBench · Reasoningtie82.783.3tie
- CritPttie12.913.4tie
FactsQwen3.7 Max
- AA-Omniscience · Non-hallucinationQwen3.7 Max by 68.55.974.4+68.5
- AA-Omniscience · AccuracyDeepSeek-V4-Pro by 11.94331.1+11.9
- SimpleQA VerifiedQwen3.7 Max by 9.646.255.8+9.6
MathQwen3.7 Max
- FrontierMath Tier 4Qwen3.7 Max by 31.82.434.1+31.8
- FrontierMath Tiers 1-3 (v2)Qwen3.7 Max by 19.345.364.6+19.3
- LiveBench · MathematicsDeepSeek-V4-Pro by 5.490.785.3+5.4
- HMMT Feb. 2026Qwen3.7 Max by 1.995.297.1+1.9
- AIME 2026tie96.797tie
- IMOAnswerBenchtie89.890tie
Following instructionsQwen3.7 Max
- LiveBench · Instruction FollowingQwen3.7 Max by 11.662.474+11.6
- IFBenchQwen3.7 Max by 476.580.5+4
Other results16 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.
- MathArena Apex (Pass@1)DeepSeek-V4-Pro by 45.790.244.5+45.7
- AA-OmniscienceQwen3.7 Max by 24.2-10.713.5+24.2
- NL2RepoQwen3.7 Max by 8.738.547.2+8.7
- Artificial Analysis Coding IndexQwen3.7 Max by 6.659.466+6.6
- HLE (with tools)DeepSeek-V4-Pro by 6.56053.5+6.5
- Toolathlon VerifiedDeepSeek-V4-Pro by 6.255.949.7+6.2
- DeepSWEQwen3.7 Max by 5.212.818+5.2
- AA Agentic IndexDeepSeek-V4-Pro by 3.827.723.9+3.8
- livebench_data_analysisDeepSeek-V4-Pro by 2.774.571.8+2.7
- MMLU-ProQwen3.7 Max by 2.187.589.6+2.1
- Terminal-Bench 2.0Qwen3.7 Max by 1.867.969.7+1.8
- livebench_languageQwen3.7 Max by 1.678.179.7+1.6
- τ²-Bench Telecom (AA run)DeepSeek-V4-Pro by 1.596.294.7+1.5
- AA Intelligencetie30.429.5tie
- LiveBenchtie73.674.3tie
- HMMT Nov. 2025tie94.495tie
Questions people ask
Which is better, DeepSeek-V4-Pro or Qwen3.7 Max?
Qwen3.7 Max wins six of the seven areas where both have results: coding, reasoning, facts, math, long documents and following instructions. DeepSeek-V4-Pro wins agents, and costs 87% less.
Which is better for coding?
Qwen3.7 Max. It wins 7 of the 10 coding tests both models report; DeepSeek-V4-Pro wins 1, and 2 are ties.
Which is cheaper?
DeepSeek-V4-Pro costs $0.43 per million input tokens and $0.87 per million output tokens; Qwen3.7 Max costs $2.50 and $7.50. That makes DeepSeek-V4-Pro about 87% cheaper for the same work.
How do you compare the two?
We use the 48 benchmark tests both models have published scores on. The verdict counts the 32 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 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.