DeepSeek-V3.2 vs DeepSeek-V3.2-Exp
Wins 4 of 6 areas
Coding · Reasoning · Long documents · Following instructions
Wins 1 of 6 areas
Facts
DeepSeek-V3.2 is the stronger all-rounder.DeepSeek-V3.2-Exp is better at facts.
Scores updated · 31 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 software50DeepSeek-V3.25 of 5 tests
- ReasoningHard problems that need careful thinking30DeepSeek-V3.23 of 3 tests
- Long documentsFinding answers in very long texts10DeepSeek-V3.21 of 1 test
- Following instructionsDoing exactly what it is asked10DeepSeek-V3.21 of 1 test
- FactsGetting facts right instead of making them up12DeepSeek-V3.2-Exp2 of 3 tests
- AgentsCarrying out multi-step tasks on its own11Even1 each
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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where DeepSeek-V3.2 pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+12.3points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+11.3points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+9.7points ahead
Where DeepSeek-V3.2-Exp pulls ahead
- Real work tasks from 44 professionsGDPVal+15.2points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+1.9points ahead
Every test, side by side
All 31 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V3.2
- SWE-bench MultilingualDeepSeek-V3.2 by 12.370.257.9+12.3
- LMArena · WebDevDeepSeek-V3.2 by 90 rating points13621272+90 rating
- SWE-bench VerifiedDeepSeek-V3.2 by 5.373.167.8+5.3
- Terminal-Bench HardDeepSeek-V3.2 by 4.535.631.1+4.5
- SciCodeDeepSeek-V3.2 by 1.238.937.7+1.2
AgentsEven
- GDPValDeepSeek-V3.2-Exp by 15.29.825+15.2
- BrowseCompDeepSeek-V3.2 by 11.351.440.1+11.3
ReasoningDeepSeek-V3.2
- Humanity's Last ExamDeepSeek-V3.2 by 9.724.614.9+9.7
- GPQA DiamondDeepSeek-V3.2 by 4.38479.7+4.3
- CritPtDeepSeek-V3.2 by 1.52.91.4+1.5
FactsDeepSeek-V3.2-Exp
- AA-Omniscience · AccuracyDeepSeek-V3.2 by 5.43327.6+5.4
- AA-Omniscience · Non-hallucinationDeepSeek-V3.2-Exp by 1.917.319.2+1.9
- Vectara HHEM hallucination ratelower is betterDeepSeek-V3.2-Exp by 16.35.3+1
Following instructionsDeepSeek-V3.2
- IFBenchDeepSeek-V3.2 by 6.660.754.1+6.6
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.
- τ²-Bench Telecom (AA run)DeepSeek-V3.2 by 56.790.633.9+56.7
- BrowseComp-ZHDeepSeek-V3.2 by 17.16547.9+17.1
- Artificial Analysis Coding IndexDeepSeek-V3.2 by 10.944.233.3+10.9
- AA Agentic IndexDeepSeek-V3.2-Exp by 10.418.328.7+10.4
- LiveCodeBenchDeepSeek-V3.2 by 9.283.374.1+9.2
- AA-OmniscienceDeepSeek-V3.2 by 8.5-22.5-31+8.5
- HMMT 2025DeepSeek-V3.2 by 6.690.283.6+6.6
- HMMT Nov. 2025DeepSeek-V3.2 by 5.89084.2+5.8
- AA IntelligenceDeepSeek-V3.2 by 4.921.516.6+4.9
- vectara_answer_rateDeepSeek-V3.2-Exp by 492.696.6+4
- AIME 2025DeepSeek-V3.2 by 3.893.189.3+3.8
- vectara_avg_summary_lengthDeepSeek-V3.2-Exp by 2.66264.6+2.6
- HMMT Feb. 2025DeepSeek-V3.2 by 2.592.590+2.5
- vectara_factual_consistencyDeepSeek-V3.2-Exp by 193.794.7+1
- MMLU-Protie8585tie
- Terminal-Benchtie37.737.7tie
Questions people ask
Which is better, DeepSeek-V3.2 or DeepSeek-V3.2-Exp?
DeepSeek-V3.2 wins four of the six areas where both have results: coding, reasoning, long documents and following instructions. DeepSeek-V3.2-Exp wins facts. They are level on agents.
Which is better for coding?
DeepSeek-V3.2. It wins 5 of the 5 coding tests both models report; DeepSeek-V3.2-Exp wins none.
Which is cheaper?
DeepSeek-V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; DeepSeek-V3.2-Exp costs $0.28 and $0.42. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 31 benchmark tests both models have published scores on. The verdict counts the 15 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.