Hy4 is the stronger all-rounder.Hy3-preview is cheaper.
Scores updated · 13 tests both models report · How we compare
Where each one wins
Tests won in each of the three 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.
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.
Hy3-preview costs 79% 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 Hy4 pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+21points ahead
- Research-level physics problemsCritPt+12points ahead
- Long professional tasks in banking, consulting and lawAA ApexAgents+11.5points ahead
Where Hy3-preview 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.
CodingHy4
- Terminal-Bench 2.1Hy4 by 2164.485.4+21
- LMArena · WebDevHy4 by 125 rating points15061631+125 rating
- SWE-bench ProHy4 by 7.857.965.7+7.8
- SWE-bench MultilingualHy4 by 7.175.882.9+7.1
Other results5 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.
- DeepSWEHy4 by 36.32864.3+36.3
- MathArena Apex (Pass@1)Hy4 by 35.538.774.2+35.5
- ToolathlonHy4 by 25.648.574.1+25.6
- NL2RepoHy4 by 13.345.658.9+13.3
- WideSearchHy4 by 7.576.483.9+7.5
Questions people ask
Which is better, Hy3-preview or Hy4?
Hy4 wins all three areas where both have results: coding, agents and reasoning. Hy3-preview wins none, but costs 79% less.
Which is better for coding?
Hy4. It wins 4 of the 4 coding tests both models report; Hy3-preview wins none.
Which is cheaper?
Hy3-preview costs $0.14 per million input tokens and $0.56 per million output tokens; Hy4 costs $0.83 and $2.50. That makes Hy3-preview about 79% cheaper for the same work.
How do you compare the two?
We use the 13 benchmark tests both models have published scores on. The verdict counts the 8 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 5 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.