DeepSeek-R1 vs MiniMax M1 80K
Wins 2 of 6 areas
Reasoning · Facts
Wins 3 of 6 areas
Agents · Long documents · Following instructions
MiniMax M1 80K wins more areas, narrowly.DeepSeek-R1 is better at facts.
Scores updated · 27 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.
- Following instructionsDoing exactly what it is asked02MiniMax M1 80K2 of 2 tests
- Long documentsFinding answers in very long texts01MiniMax M1 80K1 of 2 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own01MiniMax M1 80K1 of 1 test
- FactsGetting facts right instead of making them up20DeepSeek-R12 of 2 tests
- ReasoningHard problems that need careful thinking10DeepSeek-R11 of 3 tests · 2 ties
- CodingWriting and fixing software11Even1 each · 1 tie
Agents 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.
MiniMax M1 80K costs 54% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where MiniMax M1 80K pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+6.8points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+4points ahead
- Questions about very long textsLongBench v2+3.2points ahead
Where DeepSeek-R1 pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+8points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+3.1points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+1.2points ahead
Every test, side by side
All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- SWE-bench VerifiedMiniMax M1 80K by 6.849.256+6.8
- Terminal-Bench HardDeepSeek-R1 by 3.16.13+3.1
- SciCodetie38.337.4tie
ReasoningDeepSeek-R1
- GPQA DiamondDeepSeek-R1 by 1.170.869.7+1.1
- CritPttie0.60tie
- Humanity's Last Examtie8.58.9tie
FactsDeepSeek-R1
- AA-Omniscience · AccuracyDeepSeek-R1 by 830.522.5+8
- AA-Omniscience · Non-hallucinationDeepSeek-R1 by 1.29.78.5+1.2
Long documentsMiniMax M1 80K
- LongBench v2MiniMax M1 80K by 3.258.361.5+3.2
- AA-LCRtie57.757.7tie
Following instructionsMiniMax M1 80K
- Multi-ChallengeMiniMax M1 80K by 440.744.7+4
- IFBenchMiniMax M1 80K by 2.83941.8+2.8
Other results14 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.
- OpenAI-MRCR (128k)MiniMax M1 80K by 37.635.873.4+37.6
- τ²-Bench Telecom (AA run)MiniMax M1 80K by 22.811.434.2+22.8
- AA-OmniscienceDeepSeek-R1 by 16.3-32.2-48.5+16.3
- SimpleQADeepSeek-R1 by 11.630.118.5+11.6
- Artificial Analysis Coding IndexDeepSeek-R1 by 10.124.614.5+10.1
- LiveCodeBench (24/8~25/5)MiniMax M1 80K by 9.155.965+9.1
- ZebraLogicMiniMax M1 80K by 8.178.786.8+8.1
- AIME 2025MiniMax M1 80K by 6.97076.9+6.9
- AIME 2024MiniMax M1 80K by 6.279.886+6.2
- MMLU-ProDeepSeek-R1 by 2.98481.1+2.9
- FullStackBenchDeepSeek-R1 by 1.870.168.3+1.8
- LiveCodeBenchMiniMax M1 80K by 1.563.565+1.5
- MATH-500 (EM)tie97.396.8tie
- AA Intelligencetie11.411.7tie
Questions people ask
Which is better, DeepSeek-R1 or MiniMax M1 80K?
MiniMax M1 80K wins three of the six areas where both have results: agents, long documents and following instructions. DeepSeek-R1 wins reasoning and facts. They are level on coding.
Which is better for coding?
Neither. They win 1 coding test each of the 3 both models report, and 1 is a tie.
Which is cheaper?
DeepSeek-R1 costs $2.00 per million input tokens and $4.00 per million output tokens; MiniMax M1 80K costs $0.55 and $2.20. That makes MiniMax M1 80K about 54% cheaper for the same work.
How do you compare the two?
We use the 27 benchmark tests both models have published scores on. The verdict counts the 13 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 14 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.