Qwen3 1.7B (instruct) wins more areas, narrowly.
Scores updated · 17 tests both models report · How we compare
Where each one wins
Tests won in each of the two 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 following instructions rest on a single test each.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen3 1.7B (instruct) pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+6.8points ahead
Where Granite 4.0 1B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 17 tests both models report. The winning score is in its model's colour; marks a score checked independently.
ReasoningQwen3 1.7B (instruct)
- GPQA DiamondQwen3 1.7B (instruct) by 6.828.134.9+6.8
Other results15 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.
- GSM8KGranite 4.0 1B by 39.773.433.7+39.7
- J-MATH500Qwen3 1.7B (instruct) by 3125.456.4+31
- MATH-500 (EM)Qwen3 1.7B (instruct) by 25.644.870.4+25.6
- Avg.Qwen3 1.7B (instruct) by 17.134.151.2+17.1
- JMMLU-ProXQwen3 1.7B (instruct) by 15.515.330.8+15.5
- JMMLUQwen3 1.7B (instruct) by 13.833.947.7+13.8
- MMLU-ProQwen3 1.7B (instruct) by 9.433.542.9+9.4
- Domain AvgQwen3 1.7B (instruct) by 6.338.544.8+6.3
- BFCL v3Granite 4.0 1B by 6.152.446.3+6.1
- AIME25 no toolsQwen3 1.7B (instruct) by 63.39.3+6
- IFEvalGranite 4.0 1B by 5.979.673.7+5.9
- JHumanEval+Granite 4.0 1B by 3.651.247.6+3.6
- J-GSM8KQwen3 1.7B (instruct) by 3.242.846+3.2
- J-BFCLv3Qwen3 1.7B (instruct) by 1.950.652.5+1.9
- JGPQAQwen3 1.7B (instruct) by 1.924.426.3+1.9
Questions people ask
Which is better, Granite 4.0 1B or Qwen3 1.7B (instruct)?
Qwen3 1.7B (instruct) wins one of the two areas where both have results: reasoning. Granite 4.0 1B wins none. They are level on following instructions.
How do you compare the two?
We use the 17 benchmark tests both models have published scores on. The verdict counts the 2 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 15 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.