Model comparison
Deepseek Coder v2 vs Hy3
Hy3 is the stronger model overall, scoring 44.2 to 35.9 on the Noometry Index.
Last verified . 17 shared benchmarks.
DeepSeek
TencentSummary
- They share 17 benchmarks with published results for both. Deepseek Coder v2 scores higher in 0 categories and Hy3 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Hy3 leads 62.2 to 38.2.
Side by side
| Deepseek Coder v2 | Hy3 | |
|---|---|---|
| Provider | DeepSeek | Tencent |
| Noometry Index | 35.9 | 44.2 |
| Released | 2024-06-17 | 2026-07-06 |
| Weights | Open | Open |
| Context window | — | 262K |
| Max output | — | 128K |
| Input $ / M tokens | — | $0.13 |
| Output $ / M tokens | — | $0.53 |
| Results tracked | 24 | 19 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Hy3 leads
Deepseek Coder v2: 38.1 (#183), Hy3: 46.8 (#63)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Coding | 1251 | 1464 |
| LMArena WebDev | — | 1508 |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 59.7% | — |
| HumanEval+ | 82.3% | — |
| MBPP+ | 75.1% | — |
Reasoning Hy3 leads
Deepseek Coder v2: 23.6 (#176), Hy3: 26.1 (#136)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1447 |
| NYT Connections (extended) | — | 41.2% |
| WinoGrande | 83.7% | — |
Math Hy3 leads
Deepseek Coder v2: 34.9 (#190), Hy3: 40.1 (#93)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Math | 1241 | 1475 |
| GSM8K | 94.5% | — |
Knowledge Hy3 leads
Deepseek Coder v2: 32.3 (#212), Hy3: 40.8 (#114)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Expert | 1181 | 1460 |
| ARC (AI2) Challenge | 64.3% | — |
Multilingual Hy3 leads
Deepseek Coder v2: 36.3 (#240), Hy3: 53.5 (#65)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Non-English | 1182 | 1426 |
| LMArena Chinese | 1201 | 1493 |
| LMArena French | 1185 | 1461 |
| LMArena German | 1164 | 1439 |
| LMArena Japanese | 1126 | 1392 |
| LMArena Korean | 1104 | 1395 |
| LMArena Russian | 1188 | 1432 |
| LMArena Spanish | 1153 | 1456 |
Instruction Following Hy3 leads
Deepseek Coder v2: 61.7 (#242), Hy3: 75.1 (#70)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Instruction Following | 1180 | 1426 |
Long Context Hy3 leads
Deepseek Coder v2: 37.0 (#224), Hy3: 44.1 (#75)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Longer Query | 1219 | 1442 |
Writing & Preference Hy3 leads
Deepseek Coder v2: 38.2 (#253), Hy3: 62.2 (#81)
| Benchmark | Deepseek Coder v2 | Hy3 |
|---|---|---|
| LMArena Text | 1191 | 1439 |
| LMArena Creative Writing | 1120 | 1402 |
| LMArena Multi-Turn | 1177 | 1436 |
Frequently asked questions
Is Deepseek Coder v2 better than Hy3?
Hy3 is the stronger model overall, scoring 44.2 to 35.9 on the Noometry Index.
Is Deepseek Coder v2 or Hy3 better for coding?
Hy3 scores higher on coding benchmarks: 46.8 versus 38.1 in the Noometry coding category.
How many benchmarks do Deepseek Coder v2 and Hy3 share?
17 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and Hy3 has 19.