DeepSeek-V4-Pro vs MiMo V2.5
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
Wins 0 of 7 areas
—
DeepSeek-V4-Pro is the stronger all-rounder.MiMo V2.5 is cheaper.
Scores updated · 39 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 software50DeepSeek-V4-Pro5 of 7 tests · 2 ties
- AgentsCarrying out multi-step tasks on its own30DeepSeek-V4-Pro3 of 3 tests
- ReasoningHard problems that need careful thinking30DeepSeek-V4-Pro3 of 3 tests
- MathCompetition and research-level math20DeepSeek-V4-Pro2 of 2 tests
- FactsGetting facts right instead of making them up21DeepSeek-V4-Pro2 of 3 tests
- Long documentsFinding answers in very long texts10DeepSeek-V4-Pro1 of 1 test
- Following instructionsDoing exactly what it is asked10DeepSeek-V4-Pro1 of 1 test
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.
MiMo V2.5 costs 94% 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 DeepSeek-V4-Pro pulls ahead
- Recent programming contest problemsLiveCodeBench v6+57points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+49.8points ahead
- Short factual questions, answered correctlySimpleQA Verified+30.1points ahead
Where MiMo V2.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+62.2points ahead
Every test, side by side
All 39 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V4-Pro
- LiveCodeBench v6DeepSeek-V4-Pro by 5792.535.5+57
- SWE-bench VerifiedDeepSeek-V4-Pro by 49.880.630.8+49.8
- SciCodeDeepSeek-V4-Pro by 6.950.843.9+6.9
- Terminal-Bench HardDeepSeek-V4-Pro by 4.546.241.7+4.5
- LMArena · WebDevDeepSeek-V4-Pro by 25 rating points14631438+25 rating
- SWE-bench Protie55.456.1tie
- Terminal-Bench 2.1tie6463.7tie
AgentsDeepSeek-V4-Pro
- τ-Bench V3 · BankingDeepSeek-V4-Pro by 21.430.18.7+21.4
- Terminal-Bench 4.0DeepSeek-V4-Pro by 14.614.60+14.6
- GDPValDeepSeek-V4-Pro by 7.932.824.9+7.9
ReasoningDeepSeek-V4-Pro
- Humanity's Last ExamDeepSeek-V4-Pro by 10.337.527.2+10.3
- CritPtDeepSeek-V4-Pro by 9.212.93.7+9.2
- GPQA DiamondDeepSeek-V4-Pro by 3.988.884.9+3.9
FactsDeepSeek-V4-Pro
- AA-Omniscience · Non-hallucinationMiMo V2.5 by 62.25.968.1+62.2
- SimpleQA VerifiedDeepSeek-V4-Pro by 30.146.216.1+30.1
- AA-Omniscience · AccuracyDeepSeek-V4-Pro by 26.24316.8+26.2
MathDeepSeek-V4-Pro
- HMMT Feb. 2026DeepSeek-V4-Pro by 12.695.282.6+12.6
- AIME 2026DeepSeek-V4-Pro by 3.196.793.6+3.1
Following instructionsDeepSeek-V4-Pro
- IFBenchDeepSeek-V4-Pro by 9.476.567.1+9.4
Other results19 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.
- MMLU-ProDeepSeek-V4-Pro by 21.787.565.8+21.7
- HLE (with tools)DeepSeek-V4-Pro by 206040+20
- τ³-Bench BankingDeepSeek-V4-Pro by 17269+17
- AA IntelligenceDeepSeek-V4-Pro by 8.630.421.8+8.6
- Toolathlon VerifiedDeepSeek-V4-Pro by 6.855.949.1+6.8
- τ²-Bench Telecom (AA run)DeepSeek-V4-Pro by 5.696.290.6+5.6
- DROPDeepSeek-V4-Pro by 588.783.7+5
- TriviaQADeepSeek-V4-Pro by 4.985.680.7+4.9
- GDPval-AA v2DeepSeek-V4-Pro by 41 rating points13061265+41 rating
- MMLUDeepSeek-V4-Pro by 3.890.186.3+3.8
- MATHMiMo V2.5 by 3.264.567.7+3.2
- WinoGrandeMiMo V2.5 by 3.281.584.7+3.2
- CMMLUDeepSeek-V4-Pro by 2.690.888.2+2.6
- Terminal-Bench 2.0DeepSeek-V4-Pro by 2.167.965.8+2.1
- MMLU-ReduxDeepSeek-V4-Pro by 190.889.8+1
- AA-Omnisciencetie-10.7-9.8tie
- StrongREJECTtie98.699.3tie
- SWEBench Pro Publictie55.456.1tie
- HellaSwagtie8888.6tie
Questions people ask
Which is better, DeepSeek-V4-Pro or MiMo V2.5?
DeepSeek-V4-Pro wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. MiMo V2.5 wins none, but costs 94% less.
Which is better for coding?
DeepSeek-V4-Pro. It wins 5 of the 7 coding tests both models report; MiMo V2.5 wins none, and 2 are ties.
Which is cheaper?
DeepSeek-V4-Pro costs $1.32 per million input tokens and $3.96 per million output tokens; MiMo V2.5 costs $0.11 and $0.22. That makes MiMo V2.5 about 94% cheaper for the same work.
How do you compare the two?
We use the 39 benchmark tests both models have published scores on. The verdict counts the 20 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 19 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.