Model comparison
Hy3 vs Qwen3.5 397B-A17B
Qwen3.5 397B-A17B is the stronger model overall, scoring 46.0 to 44.2 on the Noometry Index. Hy3 costs 9.4× less per token, which makes it the better buy when Qwen3.5 397B-A17B's lead doesn't matter for your workload.
Last verified . 19 shared benchmarks.
Tencent
Alibaba (Qwen)Summary
- They share 19 benchmarks with published results for both. Hy3 scores higher in 2 categories and Qwen3.5 397B-A17B in 6 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen3.5 397B-A17B leads 53.3 to 40.8.
- The biggest single-benchmark swing is NYT Connections (extended): 41.2% for Hy3 and 58.9% for Qwen3.5 397B-A17B.
- Hy3 is cheaper at $0.0825 / $0.33 per million input/output tokens, against $0.60 / $3.60 for Qwen3.5 397B-A17B.
Side by side
| Hy3 | Qwen3.5 397B-A17B | |
|---|---|---|
| Provider | Tencent | Alibaba (Qwen) |
| Noometry Index | 44.2 | 46.0 |
| Released | 2026-07-06 | 2026-02-01 |
| Weights | Open | Open |
| Context window | 262K | 262K |
| Max output | 128K | 66K |
| Input $ / M tokens | $0.0825 | $0.60 |
| Output $ / M tokens | $0.33 | $3.60 |
| Results tracked | 19 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Hy3 leads
Hy3: 46.8 (#63), Qwen3.5 397B-A17B: 42.0 (#114)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena WebDev | 1508 | 1400 |
| LMArena Coding | 1464 | 1465 |
Agentic & Tool Use Not comparable
Hy3: —, Qwen3.5 397B-A17B: 33.3 (#53)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| APEX-Agents | — | 24.9% |
| τ²-bench Airline | — | 81.5% |
| τ²-bench Banking | — | 9.8% |
| τ²-bench Retail | — | 84.4% |
| τ²-bench Telecom | — | 97.8% |
Reasoning Qwen3.5 397B-A17B leads
Hy3: 26.1 (#136), Qwen3.5 397B-A17B: 34.5 (#70)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| NYT Connections (extended) | 41.2% | 58.9% |
| LMArena Hard Prompts | 1447 | 1448 |
| Kagi LLM Benchmark | — | 73.7% |
| Chess Puzzles | — | 13% |
| Thematic Generalization | — | 65.1% |
| Mystery Game Puzzles | — | 18% |
| DTBench | — | 87.5% |
| LMCA | — | 37.9% |
| Epoch Capabilities Index | — | 146.65 |
Math Qwen3.5 397B-A17B leads
Hy3: 40.1 (#93), Qwen3.5 397B-A17B: 46.1 (#73)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Math | 1475 | 1454 |
| FrontierMath (Tiers 1-3) | — | 31.2% |
| OTIS Mock AIME 2024-2025 | — | 88.9% |
Knowledge Qwen3.5 397B-A17B leads
Hy3: 40.8 (#114), Qwen3.5 397B-A17B: 53.3 (#58)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Expert | 1460 | 1462 |
| GPQA Diamond | — | 86.4% |
Multimodal Not comparable
Hy3: —, Qwen3.5 397B-A17B: 40.7 (#44)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Vision | — | 1263 |
Multilingual Too close to call
Hy3: 53.5 (#65), Qwen3.5 397B-A17B: 53.7 (#59)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Non-English | 1426 | 1430 |
| LMArena Chinese | 1493 | 1500 |
| LMArena French | 1461 | 1461 |
| LMArena German | 1439 | 1447 |
| LMArena Japanese | 1392 | 1426 |
| LMArena Korean | 1395 | 1384 |
| LMArena Russian | 1432 | 1429 |
| LMArena Spanish | 1456 | 1441 |
Instruction Following Too close to call
Hy3: 75.1 (#70), Qwen3.5 397B-A17B: 75.0 (#77)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Instruction Following | 1426 | 1424 |
Long Context Too close to call
Hy3: 44.1 (#75), Qwen3.5 397B-A17B: 44.1 (#74)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Longer Query | 1442 | 1442 |
Writing & Preference Too close to call
Hy3: 62.2 (#81), Qwen3.5 397B-A17B: 62.3 (#79)
| Benchmark | Hy3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Text | 1439 | 1438 |
| LMArena Creative Writing | 1402 | 1401 |
| LMArena Multi-Turn | 1436 | 1446 |
| EQ-Bench Creative Writing | — | 1478 |
Frequently asked questions
Is Hy3 better than Qwen3.5 397B-A17B?
Qwen3.5 397B-A17B is the stronger model overall, scoring 46.0 to 44.2 on the Noometry Index. Hy3 costs 9.4× less per token, which makes it the better buy when Qwen3.5 397B-A17B's lead doesn't matter for your workload.
Which is cheaper, Hy3 or Qwen3.5 397B-A17B?
Hy3 is cheaper. It lists at $0.0825 per million input tokens and $0.33 per million output tokens; Qwen3.5 397B-A17B lists at $0.60 and $3.60.
Is Hy3 or Qwen3.5 397B-A17B better for coding?
Hy3 scores higher on coding benchmarks: 46.8 versus 42.0 in the Noometry coding category.
Which has the bigger context window?
Both accept 262K tokens.
How many benchmarks do Hy3 and Qwen3.5 397B-A17B share?
19 benchmarks have published results for both models. Hy3 has 19 scored results on Noometry and Qwen3.5 397B-A17B has 36.