DeepSeek-V3.2 vs GPT-5.2
Wins 0 of 7 areas
—
Wins 6 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Following instructions
GPT-5.2 is the stronger all-rounder.DeepSeek-V3.2 is cheaper.
Scores updated · 53 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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 software15GPT-5.25 of 6 tests
- ReasoningHard problems that need careful thinking04GPT-5.24 of 4 tests
- MathCompetition and research-level math03GPT-5.23 of 3 tests
- FactsGetting facts right instead of making them up13GPT-5.23 of 4 tests
- AgentsCarrying out multi-step tasks on its own23GPT-5.23 of 5 tests
- Following instructionsDoing exactly what it is asked01GPT-5.21 of 1 test
- Long documentsFinding answers in very long texts11Even1 each
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 96% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.2 pulls ahead
Where DeepSeek-V3.2 pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+7.7points ahead
- Questions about very long textsLongBench v2+5.3points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+3.5points ahead
Every test, side by side
All 53 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.2
- SWE-bench ProGPT-5.2 by 14.315.629.9+14.3
- SciCodeGPT-5.2 by 13.238.952.1+13.2
- Terminal-Bench HardGPT-5.2 by 11.435.647+11.4
- SWE-bench VerifiedGPT-5.2 by 6.973.180+6.9
- LMArena · WebDevGPT-5.2 by 54 rating points13621416+54 rating
- SWE-bench MultilingualDeepSeek-V3.2 by 3.570.266.7+3.5
AgentsGPT-5.2
- GDPValGPT-5.2 by 38.59.848.3+38.5
- BrowseCompGPT-5.2 by 14.451.465.8+14.4
- AA ApexAgentsGPT-5.2 by 8.514.523+8.5
- τ-Bench V3 · BankingDeepSeek-V3.2 by 7.718.811.1+7.7
- MCP AtlasDeepSeek-V3.2 by 1.662.260.6+1.6
ReasoningGPT-5.2
- ARC-AGI-2GPT-5.2 by 48.9452.9+48.9
- Humanity's Last ExamGPT-5.2 by 13.124.637.7+13.1
- CritPtGPT-5.2 by 8.72.911.6+8.7
- GPQA DiamondGPT-5.2 by 6.38490.3+6.3
FactsGPT-5.2
- AA-Omniscience · AccuracyGPT-5.2 by 11.33344.3+11.3
- SimpleQA VerifiedGPT-5.2 by 9.627.537.1+9.6
- Vectara HHEM hallucination ratelower is betterDeepSeek-V3.2 by 4.56.310.8+4.5
- AA-Omniscience · Non-hallucinationGPT-5.2 by 1.517.318.8+1.5
MathGPT-5.2
- HMMT Feb. 2026GPT-5.2 by 12.984.197+12.9
- IMOAnswerBenchGPT-5.2 by 878.386.3+8
- AIME 2026GPT-5.2 by 4.194.298.3+4.1
Long documentsEven
- AA-LCRGPT-5.2 by 9.473.382.7+9.4
- LongBench v2DeepSeek-V3.2 by 5.359.854.5+5.3
Other results28 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.
- AA Agentic IndexGPT-5.2 by 41.918.360.2+41.9
- ARC-AGI-1GPT-5.2 by 29.25786.2+29.2
- AA-OmniscienceGPT-5.2 by 21.6-22.5-0.9+21.6
- SWT-benchGPT-5.2 by 18.76280.7+18.7
- frontiermath_tier_4_v1GPT-5.2 by 16.72.118.8+16.7
- PaperBenchGPT-5.2 by 16.647.163.7+16.6
- GPQA (unspecified)GPT-5.2 by 12.579.992.4+12.5
- vectara_avg_summary_lengthGPT-5.2 by 124.3 rating points62186.3+124.3 rating
- ToolathlonGPT-5.2 by 11.135.246.3+11.1
- DeepSearchQA (F1)GPT-5.2 by 10.460.971.3+10.4
- HMMT 2025GPT-5.2 by 9.290.299.4+9.2
- swe_bench_bashGPT-5.2 by 96069+9
- AA IntelligenceGPT-5.2 by 8.921.530.4+8.9
- AIME25 no toolsGPT-5.2 by 8.789.398+8.7
- Terminal-Bench 2.0GPT-5.2 by 7.646.454+7.6
- HMMT Feb. 2025GPT-5.2 by 7.592.5100+7.5
- vectara_answer_rateGPT-5.2 by 7.492.6100+7.4
- AIME 2025GPT-5.2 by 6.993.1100+6.9
- HMMT Nov. 2025GPT-5.2 by 5.89095.8+5.8
- τ²-Bench Telecom (AA run)DeepSeek-V3.2 by 5.890.684.8+5.8
- LiveCodeBenchGPT-5.2 by 5.783.389+5.7
- τ²-BenchGPT-5.2 by 5.280.385.5+5.2
- HLE (with tools)GPT-5.2 by 4.740.845.5+4.7
- Artificial Analysis Coding IndexGPT-5.2 by 4.544.248.7+4.5
- vectara_factual_consistencyDeepSeek-V3.2 by 4.593.789.2+4.5
- SWE-PerfGPT-5.2 by 2.70.93.6+2.7
- BrowseComp (context management)GPT-5.2 by 2.467.670+2.4
- MMLU-ProGPT-5.2 by 1.78586.7+1.7
Questions people ask
Which is better, DeepSeek-V3.2 or GPT-5.2?
GPT-5.2 wins six of the seven areas where both have results: coding, agents, reasoning, facts, math and following instructions. DeepSeek-V3.2 wins none, but costs 96% less. They are level on long documents.
Which is better for coding?
GPT-5.2. It wins 5 of the 6 coding tests both models report; DeepSeek-V3.2 wins 1.
Which is cheaper?
DeepSeek-V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; GPT-5.2 costs $1.75 and $14.00. That makes DeepSeek-V3.2 about 96% cheaper for the same work.
How do you compare the two?
We use the 53 benchmark tests both models have published scores on. The verdict counts the 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 28 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.