DeepSeek-V4.1-Flash vs MiMo V2.6 Pro
Wins 1 of 6 areas
Images and charts
Wins 4 of 6 areas
Coding · Agents · Reasoning · Long documents
MiMo V2.6 Pro is the stronger all-rounder.DeepSeek-V4.1-Flash is better at images and charts.
Scores updated · 22 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.
- AgentsCarrying out multi-step tasks on its own02MiMo V2.6 Pro2 of 2 tests
- ReasoningHard problems that need careful thinking02MiMo V2.6 Pro2 of 2 tests
- CodingWriting and fixing software01MiMo V2.6 Pro1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts01MiMo V2.6 Pro1 of 1 test
- Images and chartsUnderstanding pictures, charts and video10DeepSeek-V4.1-Flash1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Images and charts and long documents 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 V2.6 Pro costs 13% 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 MiMo V2.6 Pro pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+55.9points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+6.1points ahead
- Code for real scientific research problemsSciCode+5points ahead
Where DeepSeek-V4.1-Flash pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+11.5points ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingMiMo V2.6 Pro
- SciCodeMiMo V2.6 Pro by 555.960.9+5
- Terminal-Bench 2.1tie90.689.9tie
- LMArena · WebDevtie16211624tie
AgentsMiMo V2.6 Pro
- GDPValMiMo V2.6 Pro by 4.25559.2+4.2
- Terminal-Bench 4.0MiMo V2.6 Pro by 133.834.8+1
ReasoningMiMo V2.6 Pro
- Humanity's Last ExamMiMo V2.6 Pro by 6.143.349.4+6.1
- CritPtMiMo V2.6 Pro by 2.62426.6+2.6
FactsEven
- AA-Omniscience · Non-hallucinationMiMo V2.6 Pro by 55.93.559.4+55.9
- AA-Omniscience · AccuracyDeepSeek-V4.1-Flash by 11.546.434.9+11.5
Images and chartsDeepSeek-V4.1-Flash
- LMArena · VisionDeepSeek-V4.1-Flash by 13 rating points12791266+13 rating
Other results11 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-OmniscienceMiMo V2.6 Pro by 13.7-5.38.4+13.7
- HLE-DiamondDeepSeek-V4.1-Flash by 10.125.315.2+10.1
- CritPt v2 HardMiMo V2.6 Pro by 92534+9
- Terminal-Bench-Science 0.1DeepSeek-V4.1-Flash by 7.212.95.7+7.2
- AA IntelligenceMiMo V2.6 Pro by 6.839.546.3+6.8
- Program BenchMiMo V2.6 Pro by 6.220.326.5+6.2
- CyberGymMiMo V2.6 Pro by 5.988.194+5.9
- Terminal-Bench 4.0MiMo V2.6 Pro by 3.731.234.9+3.7
- DeepSWE 1.1DeepSeek-V4.1-Flash by 2.374.271.9+2.3
- GDP (Surge AI)MiMo V2.6 Pro by 217.219.2+2
- Agents' Last Examtie31.831.6tie
Questions people ask
Which is better, DeepSeek-V4.1-Flash or MiMo V2.6 Pro?
MiMo V2.6 Pro wins four of the six areas where both have results: coding, agents, reasoning and long documents. DeepSeek-V4.1-Flash wins images and charts. They are level on facts.
Which is better for coding?
MiMo V2.6 Pro. It wins 1 of the 3 coding tests both models report; DeepSeek-V4.1-Flash wins none, and 2 are ties.
Which is cheaper?
DeepSeek-V4.1-Flash costs $0.30 per million input tokens and $1.20 per million output tokens; MiMo V2.6 Pro costs $0.43 and $0.87. That makes MiMo V2.6 Pro about 13% cheaper for the same work.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 11 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.