Granite 4.2 30B vs Hy3
Which is better for the money: Granite 4.2 30B by IBM or Hy3 by Tencent? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-25.
Trade-off
Verdict
Depends on your budget
Hy3 scores 12.6 points higher (25.3 vs 12.7) but costs 1.3× as much to run the benchmark ($150 vs $111; both mid-range class). Pick Hy3 when the extra intelligence is worth the premium, otherwise Granite 4.2 30B.
| Metric | Granite 4.2 30B | Hy3 | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 12.7 | 25.3 | +99.2% Hy3 is 99.2% more intelligent |
| Total AA cost (lower is better) | $110.83 | $149.61 | +35.0% Hy3 is 35.0% more expensive |
| End-to-end time (lower is better) | 44.22s | 33.13s | −25.1% Hy3 is 25.1% faster |
| Price per 1M tokens (in / out) | $0.16 / $0.65 | $0.14 / $0.56 | |
| Input modalities | Text | Text |
Hy3 vs Granite 4.2 30B: 99.2% more intelligence, 35.0% more cost to run the AA benchmark, and 25.1% faster end-to-end.
Today's Pareto frontier
Against the frontier
- Granite 4.2 30B is off the frontier: at the same or lower cost, MiMo-V2.6-Flash scores 37.90 for $110.
- Hy3 is off the frontier: at the same or lower cost, GPT-6 Luna (Max) scores 38.10 for $122.
Granite 4.2 30BHy3Today's Pareto frontierOpen-weights frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/granite-4-2-30b/hy3 · Markdown: /api/v1/compare/granite-4-2-30b/hy3?format=md · llms.txt · agent skill