DeepSeek-V3.2 vs Gemini 2.5 Pro
Wins 5 of 6 areas
Coding · Agents · Reasoning · Long documents · Following instructions
Wins 1 of 6 areas
Facts
DeepSeek-V3.2 is the stronger all-rounder.Gemini 2.5 Pro is better at facts.
Scores updated · 45 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 software41DeepSeek-V3.24 of 5 tests
- AgentsCarrying out multi-step tasks on its own30DeepSeek-V3.23 of 3 tests
- Long documentsFinding answers in very long texts20DeepSeek-V3.22 of 2 tests
- ReasoningHard problems that need careful thinking10DeepSeek-V3.21 of 4 tests · 3 ties
- Following instructionsDoing exactly what it is asked10DeepSeek-V3.21 of 1 test
- FactsGetting facts right instead of making them up12Gemini 2.5 Pro2 of 4 tests · 1 tie
Following instructions 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-V3.2 costs 94% 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-V3.2 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+41.5points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+18.3points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+12points ahead
Where Gemini 2.5 Pro pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+28.5points ahead
- Code for real scientific research problemsSciCode+7.4points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+6.1points ahead
Every test, side by side
All 45 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V3.2
- Terminal-Bench 2.1DeepSeek-V3.2 by 18.346.828.5+18.3
- LMArena · WebDevDeepSeek-V3.2 by 138 rating points13621224+138 rating
- SWE-bench VerifiedDeepSeek-V3.2 by 9.973.163.2+9.9
- Terminal-Bench HardDeepSeek-V3.2 by 9.135.626.5+9.1
- SciCodeGemini 2.5 Pro by 7.438.946.3+7.4
AgentsDeepSeek-V3.2
- BrowseCompDeepSeek-V3.2 by 41.551.49.9+41.5
- GDPValDeepSeek-V3.2 by 9.89.80+9.8
- τ-Bench V3 · BankingDeepSeek-V3.2 by 9.118.89.7+9.1
ReasoningDeepSeek-V3.2
- Humanity's Last ExamDeepSeek-V3.2 by 6.624.618+6.6
- ARC-AGI-2tie44.9tie
- GPQA Diamondtie8483.6tie
- CritPttie2.92.6tie
FactsGemini 2.5 Pro
- SimpleQA VerifiedGemini 2.5 Pro by 28.527.556+28.5
- AA-Omniscience · Non-hallucinationDeepSeek-V3.2 by 8.217.39.1+8.2
- AA-Omniscience · AccuracyGemini 2.5 Pro by 6.13339+6.1
- Vectara HHEM hallucination ratelower is bettertie6.37tie
Long documentsDeepSeek-V3.2
- AA-LCRDeepSeek-V3.2 by 4.373.369+4.3
- LongBench v2DeepSeek-V3.2 by 159.858.8+1
Following instructionsDeepSeek-V3.2
- IFBenchDeepSeek-V3.2 by 1260.748.7+12
Other results26 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.
- MRCRGemini 2.5 Pro by 37.555.593+37.5
- τ²-Bench Telecom (AA run)DeepSeek-V3.2 by 36.590.654.1+36.5
- BrowseComp-ZHDeepSeek-V3.2 by 32.86532.2+32.8
- τ²-BenchDeepSeek-V3.2 by 26.380.354+26.3
- HMMT 2025DeepSeek-V3.2 by 24.590.265.7+24.5
- ARC-AGI-1DeepSeek-V3.2 by 205737+20
- FinSearchComp-globalGemini 2.5 Pro by 16.426.242.6+16.4
- AA Agentic IndexDeepSeek-V3.2 by 14.818.33.5+14.8
- HLE (with tools)DeepSeek-V3.2 by 12.440.828.4+12.4
- Terminal-BenchDeepSeek-V3.2 by 12.437.725.3+12.4
- HMMT Feb. 2025DeepSeek-V3.2 by 11.792.580.8+11.7
- AIME 2025DeepSeek-V3.2 by 10.193.183+10.1
- HMMT Nov. 2025DeepSeek-V3.2 by 109080+10
- LiveCodeBenchDeepSeek-V3.2 by 9.183.374.2+9.1
- AA IntelligenceDeepSeek-V3.2 by 6.521.515+6.5
- vectara_answer_rateGemini 2.5 Pro by 6.592.699.1+6.5
- AA-OmniscienceGemini 2.5 Pro by 6.2-22.5-16.4+6.2
- theagentcompanyGemini 2.5 Pro by 5.33439.3+5.3
- vectara_avg_summary_lengthGemini 2.5 Pro by 44.4 rating points62106.4+44.4 rating
- GAIA (text only)DeepSeek-V3.2 by 3.363.560.2+3.3
- Artificial Analysis Coding IndexGemini 2.5 Pro by 2.544.246.7+2.5
- frontiermath_tier_4_v1Gemini 2.5 Pro by 2.12.14.2+2.1
- ArtifactsBenchGemini 2.5 Pro by 1.955.857.7+1.9
- MMLU-ProGemini 2.5 Pro by 18586+1
- vectara_factual_consistencytie93.793tie
- xbench-DeepSearchtie55.756tie
Questions people ask
Which is better, DeepSeek-V3.2 or Gemini 2.5 Pro?
DeepSeek-V3.2 wins five of the six areas where both have results: coding, agents, reasoning, long documents and following instructions. Gemini 2.5 Pro wins facts.
Which is better for coding?
DeepSeek-V3.2. It wins 4 of the 5 coding tests both models report; Gemini 2.5 Pro wins 1.
Which is cheaper?
DeepSeek-V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; Gemini 2.5 Pro costs $1.25 and $10.00. That makes DeepSeek-V3.2 about 94% cheaper for the same work.
How do you compare the two?
We use the 45 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 26 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.