DeepSeek-V3 vs MAI-Thinking-1
Wins 0 of 4 areas
—
Wins 4 of 4 areas
Coding · Reasoning · Long documents · Following instructions
MAI-Thinking-1 is the stronger all-rounder.
Scores updated · 13 tests both models report · How we compare
Where each one wins
Tests won in each of the four 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.
Reasoning 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where MAI-Thinking-1 pulls ahead
- Recent programming contest problemsLiveCodeBench v6+40.8points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+31.5points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+28points ahead
Where DeepSeek-V3 pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 13 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingMAI-Thinking-1
- LiveCodeBench v6MAI-Thinking-1 by 40.846.987.7+40.8
- SWE-bench VerifiedMAI-Thinking-1 by 31.54273.5+31.5
Long documentsMAI-Thinking-1
- LongBench v2MAI-Thinking-1 by 12.348.761+12.3
Following instructionsMAI-Thinking-1
- IFBenchMAI-Thinking-1 by 284169+28
- Multi-ChallengeMAI-Thinking-1 by 21.631.453+21.6
Other results7 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.
Questions people ask
Which is better, DeepSeek-V3 or MAI-Thinking-1?
MAI-Thinking-1 wins all four areas where both have results: coding, reasoning, long documents and following instructions. DeepSeek-V3 wins none.
Which is better for coding?
MAI-Thinking-1. It wins 2 of the 2 coding tests both models report; DeepSeek-V3 wins none.
How do you compare the two?
We use the 13 benchmark tests both models have published scores on. The verdict counts the 6 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 7 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.