DeepSeek-V3.1 vs Gemini 3 Pro
Wins 1 of 6 areas
Facts
Wins 5 of 6 areas
Coding · Agents · Reasoning · Long documents · Following instructions
Gemini 3 Pro is the stronger all-rounder.DeepSeek-V3.1 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 software04Gemini 3 Pro4 of 4 tests
- ReasoningHard problems that need careful thinking04Gemini 3 Pro4 of 4 tests
- AgentsCarrying out multi-step tasks on its own02Gemini 3 Pro2 of 2 tests
- Following instructionsDoing exactly what it is asked02Gemini 3 Pro2 of 2 tests
- Long documentsFinding answers in very long texts01Gemini 3 Pro1 of 1 test
- FactsGetting facts right instead of making them up21DeepSeek-V3.12 of 3 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-V3.1 costs 84% 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 Gemini 3 Pro pulls ahead
- Common-sense trick questionsSimpleBench+36.4points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+29.2points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+28.9points ahead
Where DeepSeek-V3.1 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+9points 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.
CodingGemini 3 Pro
- SciCodeGemini 3 Pro by 1739.156.1+17
- Terminal-Bench HardGemini 3 Pro by 16.72541.7+16.7
- SWE-bench MultilingualGemini 3 Pro by 10.554.565+10.5
- SWE-bench VerifiedGemini 3 Pro by 10.26676.2+10.2
AgentsGemini 3 Pro
- BrowseCompGemini 3 Pro by 29.23059.2+29.2
- GDPValGemini 3 Pro by 28.65.634.2+28.6
ReasoningGemini 3 Pro
- SimpleBenchGemini 3 Pro by 36.44076.4+36.4
- Humanity's Last ExamGemini 3 Pro by 25.414.339.7+25.4
- GPQA DiamondGemini 3 Pro by 12.977.990.8+12.9
- CritPtGemini 3 Pro by 7.129.1+7.1
FactsDeepSeek-V3.1
- AA-Omniscience · AccuracyGemini 3 Pro by 26.82955.8+26.8
- AA-Omniscience · Non-hallucinationDeepSeek-V3.1 by 917.58.5+9
- Vectara HHEM hallucination ratelower is betterDeepSeek-V3.1 by 8.15.513.6+8.1
Following instructionsGemini 3 Pro
- IFBenchGemini 3 Pro by 28.941.570.4+28.9
- Multi-ChallengeGemini 3 Pro by 19.646.165.7+19.6
Other results13 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.
- AIME 2025Gemini 3 Pro by 50.249.8100+50.2
- τ²-Bench Telecom (AA run)Gemini 3 Pro by 49.737.487.1+49.7
- AA-OmniscienceGemini 3 Pro by 44.9-29.615.3+44.9
- LiveCodeBenchGemini 3 Pro by 35.656.492+35.6
- AA Agentic IndexGemini 3 Pro by 33.118.952+33.1
- SimpleQADeepSeek-V3.1 by 21.393.472.1+21.3
- Artificial Analysis Coding IndexGemini 3 Pro by 16.829.746.5+16.8
- AA IntelligenceGemini 3 Pro by 14.513.528+14.5
- HMMT Feb. 2025Gemini 3 Pro by 11.785.897.5+11.7
- vectara_factual_consistencyDeepSeek-V3.1 by 8.194.586.4+8.1
- MMLU-ProGemini 3 Pro by 6.483.790.1+6.4
- vectara_answer_rateGemini 3 Pro by 4.994.599.4+4.9
- vectara_avg_summary_lengthGemini 3 Pro by 38.2 rating points63.7101.9+38.2 rating
Questions people ask
Which is better, DeepSeek-V3.1 or Gemini 3 Pro?
Gemini 3 Pro wins five of the six areas where both have results: coding, agents, reasoning, long documents and following instructions. DeepSeek-V3.1 wins facts, and costs 84% less.
Which is better for coding?
Gemini 3 Pro. It wins 4 of the 4 coding tests both models report; DeepSeek-V3.1 wins none.
Which is cheaper?
DeepSeek-V3.1 costs $0.56 per million input tokens and $1.68 per million output tokens; Gemini 3 Pro costs $2.00 and $12.00. That makes DeepSeek-V3.1 about 84% 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 16 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.