DeepSeek-V3 is the stronger all-rounder.
Scores updated · 25 tests both models report · How we compare
Where each one wins
Tests won in each of the two 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.
Coding 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-V3 costs 97% 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 pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+17.5points ahead
- Code for real scientific research problemsSciCode+7.1points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+1.6points ahead
Where GPT-4 Turbo pulls ahead
- Common-sense trick questionsSimpleBench+6.2points ahead
Every test, side by side
All 25 tests both models report. The winning score is in its model's colour; marks a score checked independently.
ReasoningDeepSeek-V3
- GPQA DiamondDeepSeek-V3 by 17.565.548+17.5
- SimpleBenchGPT-4 Turbo by 6.218.925.1+6.2
- Humanity's Last ExamDeepSeek-V3 by 1.64.73.1+1.6
Other results21 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.
- HarmBenchGPT-4 Turbo by 40.149.789.8+40.1
- AIR-Bench 2024GPT-4 Turbo by 31.140.871.9+31.1
- livecodebench_mediumDeepSeek-V3 by 22.853.530.7+22.8
- HumanEvalGPT-4 Turbo by 21.965.287.1+21.9
- MMLUDeepSeek-V3 by 17.488.571.1+17.4
- LiveCodeBenchDeepSeek-V3 by 11.949.237.3+11.9
- GSM8KDeepSeek-V3 by 11.69482.4+11.6
- livecodebench_hardDeepSeek-V3 by 10.214.24+10.2
- MGSMGPT-4 Turbo by 8.779.888.5+8.7
- MATHDeepSeek-V3 by 6.990.283.3+6.9
- DROPDeepSeek-V3 by 5.691.686+5.6
- simple_safety_testsGPT-4 Turbo by 3.795.399+3.7
- NarrativeQADeepSeek-V3 by 3.579.676.1+3.5
- AA IntelligenceDeepSeek-V3 by 2.79.77+2.7
- anthropic_red_teamGPT-4 Turbo by 2.697.199.7+2.6
- bbqDeepSeek-V3 by 2.696.794.1+2.6
- livecodebench_easyDeepSeek-V3 by 1.983.381.4+1.9
- OpenBookQAGPT-4 Turbo by 1.695.497+1.6
- Artificial Analysis Coding IndexDeepSeek-V3 by 1.52321.5+1.5
- naturalquestions_closedbookGPT-4 Turbo by 1.546.748.2+1.5
- XSTesttie97.197.7tie
Questions people ask
Which is better, DeepSeek-V3 or GPT-4 Turbo?
DeepSeek-V3 wins all two areas where both have results: coding and reasoning. GPT-4 Turbo wins none.
Which is better for coding?
DeepSeek-V3. It wins the one coding test both models report.
Which is cheaper?
DeepSeek-V3 costs $0.24 per million input tokens and $0.90 per million output tokens; GPT-4 Turbo costs $10.00 and $30.00. That makes DeepSeek-V3 about 97% cheaper for the same work.
How do you compare the two?
We use the 25 benchmark tests both models have published scores on. The verdict counts the 4 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 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.