DeepSeek-V3.2 vs MiMo V2 Flash
Wins 4 of 7 areas
Coding · Agents · Facts · Long documents
Wins 2 of 7 areas
Math · Following instructions
DeepSeek-V3.2 wins more areas, narrowly.MiMo V2 Flash is cheaper and better at math.
Scores updated · 41 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.
- FactsGetting facts right instead of making them up20DeepSeek-V3.22 of 2 tests
- CodingWriting and fixing software32DeepSeek-V3.23 of 7 tests · 2 ties
- AgentsCarrying out multi-step tasks on its own21DeepSeek-V3.22 of 3 tests
- Long documentsFinding answers in very long texts10DeepSeek-V3.21 of 2 tests · 1 tie
- MathCompetition and research-level math01MiMo V2 Flash1 of 1 test
- Following instructionsDoing exactly what it is asked01MiMo V2 Flash1 of 1 test
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
Math 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.
MiMo V2 Flash costs 43% 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
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+15.7points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+10.7points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+8.2points ahead
Where MiMo V2 Flash pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+15points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+6.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+3.5points ahead
Every test, side by side
All 41 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V3.2
- Terminal-Bench 2.1MiMo V2 Flash by 1546.861.8+15
- Terminal-Bench HardDeepSeek-V3.2 by 7.635.628+7.6
- LMArena · WebDevDeepSeek-V3.2 by 31 rating points13621331+31 rating
- LiveCodeBench v6DeepSeek-V3.2 by 2.783.380.6+2.7
- SWE-bench MultilingualMiMo V2 Flash by 1.570.271.7+1.5
- SciCodetie38.939.4tie
- SWE-bench Verifiedtie73.173.4tie
AgentsDeepSeek-V3.2
- τ-Bench V3 · BankingDeepSeek-V3.2 by 15.718.83.1+15.7
- BrowseCompMiMo V2 Flash by 6.951.458.3+6.9
- GDPValDeepSeek-V3.2 by 3.29.86.6+3.2
ReasoningEven
- Humanity's Last ExamDeepSeek-V3.2 by 1.824.622.8+1.8
- CritPtMiMo V2 Flash by 1.42.94.3+1.4
- GPQA Diamondtie8484.6tie
FactsDeepSeek-V3.2
- AA-Omniscience · Non-hallucinationDeepSeek-V3.2 by 10.717.36.6+10.7
- AA-Omniscience · AccuracyDeepSeek-V3.2 by 8.23324.8+8.2
Long documentsDeepSeek-V3.2
- AA-LCRDeepSeek-V3.2 by 2.673.370.7+2.6
- LongBench v2tie59.860.6tie
Following instructionsMiMo V2 Flash
- IFBenchMiMo V2 Flash by 3.560.764.2+3.5
Other results22 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-OmniscienceDeepSeek-V3.2 by 22.8-22.5-45.4+22.8
- BrowseComp-ZHDeepSeek-V3.2 by 13.86551.2+13.8
- xbench-DeepSearchDeepSeek-V3.2 by 11.755.744+11.7
- xbench-DeepSearch (2025.10)DeepSeek-V3.2 by 11.755.744+11.7
- MRCRDeepSeek-V3.2 by 9.855.545.7+9.8
- browsecomp_with_context_managerDeepSeek-V3.2 by 9.367.658.3+9.3
- HMMT Feb. 2025DeepSeek-V3.2 by 8.192.584.4+8.1
- Terminal-Bench 2.0DeepSeek-V3.2 by 7.946.438.5+7.9
- Terminal-BenchDeepSeek-V3.2 by 7.237.730.5+7.2
- AA Agentic IndexDeepSeek-V3.2 by 6.318.312+6.3
- HMMT 2025DeepSeek-V3.2 by 5.890.284.4+5.8
- Artificial Analysis Coding IndexMiMo V2 Flash by 5.644.249.8+5.6
- τ²-Bench Telecom (AA run)MiMo V2 Flash by 4.490.695+4.4
- GAIA (no file)MiMo V2 Flash by 3.175.178.2+3.1
- Arena-Hard (Creative Writing)DeepSeek-V3.2 by 2.688.886.2+2.6
- ResearchRubricsDeepSeek-V3.2 by 1.555.854.3+1.5
- AIME 2025MiMo V2 Flash by 193.194.1+1
- HMMT Nov. 2025MiMo V2 Flash by 19091+1
- AA Intelligencetie21.520.8tie
- Arena-Hard (Hard Prompt)tie53.454.1tie
- MMLU-Protie8584.9tie
- τ²-Benchtie80.380.3tie
Questions people ask
Which is better, DeepSeek-V3.2 or MiMo V2 Flash?
DeepSeek-V3.2 wins four of the seven areas where both have results: coding, agents, facts and long documents. MiMo V2 Flash wins math and following instructions, and costs 43% less. They are level on reasoning.
Which is better for coding?
DeepSeek-V3.2. It wins 3 of the 7 coding tests both models report; MiMo V2 Flash wins 2, and 2 are ties.
Which is cheaper?
DeepSeek-V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; MiMo V2 Flash costs $0.10 and $0.30. That makes MiMo V2 Flash about 43% cheaper for the same work.
How do you compare the two?
We use the 41 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 22 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.