Granite 4.0 350M vs Qwen3 235B A22B Instruct 2507
Wins 0 of 6 areas
—
Wins 6 of 6 areas
Coding · Agents · Reasoning · Facts · Long documents · Following instructions
Qwen3 235B A22B Instruct 2507 is the stronger all-rounder.
Scores updated · 18 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.
- ReasoningHard problems that need careful thinking02Qwen3 235B A22B Instruct 25072 of 3 tests · 1 tie
- CodingWriting and fixing software02Qwen3 235B A22B Instruct 25072 of 2 tests
- FactsGetting facts right instead of making them up02Qwen3 235B A22B Instruct 25072 of 2 tests
- AgentsCarrying out multi-step tasks on its own01Qwen3 235B A22B Instruct 25071 of 1 test
- Long documentsFinding answers in very long texts01Qwen3 235B A22B Instruct 25071 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen3 235B A22B Instruct 25071 of 1 test
Agents, long documents and following instructions 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 Qwen3 235B A22B Instruct 2507 pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+49.6points ahead
- Code for real scientific research problemsSciCode+34.3points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+28.5points ahead
Where Granite 4.0 350M pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 18 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 235B A22B Instruct 2507
- SciCodeQwen3 235B A22B Instruct 2507 by 34.31.736+34.3
- Terminal-Bench HardQwen3 235B A22B Instruct 2507 by 15.2015.2+15.2
AgentsQwen3 235B A22B Instruct 2507
- GDPValQwen3 235B A22B Instruct 2507 by 14.1014.1+14.1
ReasoningQwen3 235B A22B Instruct 2507
- GPQA DiamondQwen3 235B A22B Instruct 2507 by 49.625.775.3+49.6
- Humanity's Last ExamQwen3 235B A22B Instruct 2507 by 4.76.411.1+4.7
- CritPttie00tie
FactsQwen3 235B A22B Instruct 2507
- AA-Omniscience · AccuracyQwen3 235B A22B Instruct 2507 by 14.93.818.7+14.9
- AA-Omniscience · Non-hallucinationQwen3 235B A22B Instruct 2507 by 10.911.922.8+10.9
Long documentsQwen3 235B A22B Instruct 2507
- AA-LCRQwen3 235B A22B Instruct 2507 by 33.9033.9+33.9
Following instructionsQwen3 235B A22B Instruct 2507
- IFBenchQwen3 235B A22B Instruct 2507 by 28.517.646.1+28.5
Other results8 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.
- MMLU-ProQwen3 235B A22B Instruct 2507 by 70.212.883+70.2
- τ²-Bench (Retail)Qwen3 235B A22B Instruct 2507 by 65.26.171.3+65.2
- AA-OmniscienceQwen3 235B A22B Instruct 2507 by 36.9-80.9-44+36.9
- IFEvalQwen3 235B A22B Instruct 2507 by 35.253.588.7+35.2
- BFCL v3Qwen3 235B A22B Instruct 2507 by 31.339.670.9+31.3
- τ²-Bench Telecom (AA run)Qwen3 235B A22B Instruct 2507 by 18.714.633.3+18.7
- AA Agentic IndexQwen3 235B A22B Instruct 2507 by 17.94.922.8+17.9
- AA IntelligenceQwen3 235B A22B Instruct 2507 by 7.24.812+7.2
Questions people ask
Which is better, Granite 4.0 350M or Qwen3 235B A22B Instruct 2507?
Qwen3 235B A22B Instruct 2507 wins all six areas where both have results: coding, agents, reasoning, facts, long documents and following instructions. Granite 4.0 350M wins none.
Which is better for coding?
Qwen3 235B A22B Instruct 2507. It wins 2 of the 2 coding tests both models report; Granite 4.0 350M wins none.
How do you compare the two?
We use the 18 benchmark tests both models have published scores on. The verdict counts the 10 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 8 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.