Model comparison
Grok-3 mini vs Hy4 preview
Hy4 preview is the stronger model overall, scoring 44.6 to 41.0 on the Noometry Index.
Last verified . 0 shared benchmarks.
TencentSummary
- The widest gap is in reasoning, where Hy4 preview leads 30.6 to 13.8.
- Hy4 preview has downloadable open weights; the other is API-only.
Side by side
| Grok-3 mini | Hy4 preview | |
|---|---|---|
| Provider | xAI | Tencent |
| Noometry Index | 41.0 | 44.6 |
| Released | 2025-04-09 | 2026-08-28 |
| Weights | Proprietary | Open |
| Context window | — | 1.05M |
| Max output | — | 64K |
| Input $ / M tokens | — | $0.83 |
| Output $ / M tokens | — | $2.50 |
| Results tracked | 35 | 3 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Hy4 preview leads
Grok-3 mini: 40.4 (#126), Hy4 preview: 51.4 (#36)
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| Aider Polyglot | 49.3% | — |
| LMArena WebDev | — | 1633 |
| WeirdML | 42.6% | — |
| LMArena Coding | 1380 | — |
Reasoning Hy4 preview leads
Grok-3 mini: 13.8 (#323), Hy4 preview: 30.6 (#75)
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| ARC-AGI-2 | 0.4% | — |
| Kagi LLM Benchmark | 61.3% | — |
| NYT Connections (extended) | — | 68.2% |
| ARC-AGI-1 | 16.5% | — |
| LMArena Hard Prompts | 1374 | — |
| Epoch Capabilities Index | 140.35 | — |
Math Hy4 preview leads
Grok-3 mini: 40.3 (#78), Hy4 preview: 54.5 (#37)
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 77.8% | — |
| ProofBench | — | 75% |
| Omni-MATH | 31.8% | — |
| LMArena Math | 1386 | — |
| MATH Level 5 | 90.9% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
Knowledge Not comparable
Grok-3 mini: 46.3 (#75), Hy4 preview: —
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| GPQA Diamond | 76.3% | — |
| MMLU-Pro | 79.9% | — |
| Confabulations | 10.8% | — |
| GPQA (HELM) | 67.5% | — |
| LMArena Expert | 1395 | — |
Multilingual Not comparable
Grok-3 mini: 48.3 (#141), Hy4 preview: —
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| LMArena Non-English | 1352 | — |
| LMArena Chinese | 1387 | — |
| LMArena French | 1357 | — |
| LMArena German | 1350 | — |
| LMArena Japanese | 1343 | — |
| LMArena Korean | 1336 | — |
| LMArena Russian | 1352 | — |
| LMArena Spanish | 1379 | — |
Instruction Following Not comparable
Grok-3 mini: 78.7 (#8), Hy4 preview: —
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| IFEval | 95.1% | — |
| LMArena Instruction Following | 1356 | — |
Long Context Not comparable
Grok-3 mini: 41.0 (#142), Hy4 preview: —
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| Fiction.LiveBench | 66.7% | — |
| LMArena Longer Query | 1372 | — |
Writing & Preference Not comparable
Grok-3 mini: 52.7 (#163), Hy4 preview: —
| Benchmark | Grok-3 mini | Hy4 preview |
|---|---|---|
| LMArena Text | 1370 | — |
| LMArena Creative Writing | 1342 | — |
| Short-Story Creative Writing | 73.5% | — |
| WildBench | 65.1% | — |
| LMArena Multi-Turn | 1355 | — |
Frequently asked questions
Is Grok-3 mini better than Hy4 preview?
Hy4 preview is the stronger model overall, scoring 44.6 to 41.0 on the Noometry Index.
Is Grok-3 mini or Hy4 preview better for coding?
Hy4 preview scores higher on coding benchmarks: 51.4 versus 40.4 in the Noometry coding category.
How many benchmarks do Grok-3 mini and Hy4 preview share?
0 benchmarks have published results for both models. Grok-3 mini has 35 scored results on Noometry and Hy4 preview has 3.