DeepSeek-V3 vs DeepSeek-V4-Pro
Wins 1 of 6 areas
Facts
Wins 5 of 6 areas
Coding · Agents · Reasoning · Long documents · Following instructions
DeepSeek-V4-Pro is the stronger all-rounder.DeepSeek-V3 is cheaper and better at facts.
Scores updated · 29 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 software05DeepSeek-V4-Pro5 of 5 tests
- ReasoningHard problems that need careful thinking04DeepSeek-V4-Pro4 of 4 tests
- AgentsCarrying out multi-step tasks on its own03DeepSeek-V4-Pro3 of 3 tests
- Long documentsFinding answers in very long texts01DeepSeek-V4-Pro1 of 1 test
- Following instructionsDoing exactly what it is asked01DeepSeek-V4-Pro1 of 1 test
- FactsGetting facts right instead of making them up21DeepSeek-V32 of 3 tests
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 costs 13% 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-V4-Pro pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+47.1points ahead
- Recent programming contest problemsLiveCodeBench v6+45.6points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+38.6points ahead
Where DeepSeek-V3 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+8.2points ahead
Every test, side by side
All 29 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V4-Pro
- Terminal-Bench 2.1DeepSeek-V4-Pro by 47.116.964+47.1
- LiveCodeBench v6DeepSeek-V4-Pro by 45.646.992.5+45.6
- SWE-bench VerifiedDeepSeek-V4-Pro by 38.64280.6+38.6
- Terminal-Bench HardDeepSeek-V4-Pro by 3115.246.2+31
- SciCodeDeepSeek-V4-Pro by 11.83950.8+11.8
AgentsDeepSeek-V4-Pro
- GDPValDeepSeek-V4-Pro by 32.9032.9+32.9
- τ-Bench V3 · BankingDeepSeek-V4-Pro by 25.44.730.1+25.4
- Terminal-Bench 4.0DeepSeek-V4-Pro by 14.6014.6+14.6
ReasoningDeepSeek-V4-Pro
- Humanity's Last ExamDeepSeek-V4-Pro by 32.84.737.5+32.8
- SimpleBenchDeepSeek-V4-Pro by 3218.950.9+32
- GPQA DiamondDeepSeek-V4-Pro by 23.365.588.8+23.3
- CritPtDeepSeek-V4-Pro by 12.9012.9+12.9
FactsDeepSeek-V3
- AA-Omniscience · AccuracyDeepSeek-V4-Pro by 17.525.443+17.5
- AA-Omniscience · Non-hallucinationDeepSeek-V3 by 8.214.15.9+8.2
- Vectara HHEM hallucination ratelower is betterDeepSeek-V3 by 2.56.18.6+2.5
Following instructionsDeepSeek-V4-Pro
- IFBenchDeepSeek-V4-Pro by 35.54176.5+35.5
Other results12 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.
- Codeforces (Rating)DeepSeek-V4-Pro by 2214 rating points11343348+2214 rating
- τ²-Bench Telecom (AA run)DeepSeek-V4-Pro by 49.147.196.2+49.1
- LiveCodeBenchDeepSeek-V4-Pro by 44.349.293.5+44.3
- Artificial Analysis Coding IndexDeepSeek-V4-Pro by 36.42359.4+36.4
- SimpleQADeepSeek-V4-Pro by 3324.957.9+33
- AA-OmniscienceDeepSeek-V4-Pro by 30-40.7-10.7+30
- AA IntelligenceDeepSeek-V4-Pro by 20.79.730.4+20.7
- MMLU-ProDeepSeek-V4-Pro by 11.675.987.5+11.6
- vectara_avg_summary_lengthDeepSeek-V4-Pro by 72.1 rating points81.7153.8+72.1 rating
- vectara_factual_consistencyDeepSeek-V3 by 2.593.991.4+2.5
- LiveBenchDeepSeek-V4-Pro by 1.272.473.6+1.2
- vectara_answer_ratetie97.597.2tie
Questions people ask
Which is better, DeepSeek-V3 or DeepSeek-V4-Pro?
DeepSeek-V4-Pro wins five of the six areas where both have results: coding, agents, reasoning, long documents and following instructions. DeepSeek-V3 wins facts, and costs 13% less.
Which is better for coding?
DeepSeek-V4-Pro. It wins 5 of the 5 coding tests both models report; DeepSeek-V3 wins none.
Which is cheaper?
DeepSeek-V3 costs $0.24 per million input tokens and $0.90 per million output tokens; DeepSeek-V4-Pro costs $0.43 and $0.87. That makes DeepSeek-V3 about 13% cheaper for the same work.
How do you compare the two?
We use the 29 benchmark tests both models have published scores on. The verdict counts the 17 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 12 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.