Model comparison
Grok 3 vs Qwen3.5 397B-A17B
Qwen3.5 397B-A17B is the stronger model overall, scoring 45.7 to 39.6 on the Noometry Index.
Last verified . 22 shared benchmarks.
Alibaba (Qwen)Summary
- They share 22 benchmarks with published results for both. Grok 3 scores higher in 1 category and Qwen3.5 397B-A17B in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.5 397B-A17B leads 33.6 to 13.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 55.6% for Grok 3 and 88.9% for Qwen3.5 397B-A17B.
- Qwen3.5 397B-A17B has downloadable open weights; the other is API-only.
Side by side
| Grok 3 | Qwen3.5 397B-A17B | |
|---|---|---|
| Provider | xAI | Alibaba (Qwen) |
| Noometry Index | 39.6 | 45.7 |
| Released | 2025-04-09 | 2026-02-01 |
| Weights | Proprietary | Open |
| Context window | — | 262K |
| Max output | — | 66K |
| Input $ / M tokens | — | $0.60 |
| Output $ / M tokens | — | $3.60 |
| Results tracked | 40 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Grok 3: 41.5 (#113), Qwen3.5 397B-A17B: 41.7 (#108)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Coding | 1432 | 1464 |
| Aider Polyglot | 53.3% | — |
| LMArena WebDev | — | 1400 |
| WeirdML | 37.2% | — |
Agentic & Tool Use Qwen3.5 397B-A17B leads
Grok 3: 30.2 (#69), Qwen3.5 397B-A17B: 33.2 (#51)
| Benchmark | Grok 3 | 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% |
| BALROG | 29.5% | — |
Reasoning Qwen3.5 397B-A17B leads
Grok 3: 13.8 (#322), Qwen3.5 397B-A17B: 33.6 (#62)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| Kagi LLM Benchmark | 61.3% | 73.7% |
| LMArena Hard Prompts | 1433 | 1448 |
| Epoch Capabilities Index | 138.34 | 146.64 |
| ARC-AGI-2 | 0% | — |
| SimpleBench | 36.1% | — |
| NYT Connections (extended) | — | 58.9% |
| ARC-AGI-1 | 5.5% | — |
| Chess Puzzles | — | 13% |
| Thematic Generalization | — | 65.1% |
| Mystery Game Puzzles | — | 18% |
| DTBench | — | 87.5% |
| LMCA | — | 37.9% |
Math Qwen3.5 397B-A17B leads
Grok 3: 36.2 (#122), Qwen3.5 397B-A17B: 44.9 (#64)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 55.6% | 88.9% |
| LMArena Math | 1390 | 1455 |
| FrontierMath (Tiers 1-3) | — | 31.2% |
| Omni-MATH | 46.4% | — |
| MATH Level 5 | 88.7% | — |
| FrontierMath (Feb 2025 set) | 3.8% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Qwen3.5 397B-A17B leads
Grok 3: 46.0 (#76), Qwen3.5 397B-A17B: 53.3 (#53)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| GPQA Diamond | 75.8% | 86.4% |
| LMArena Expert | 1420 | 1461 |
| MMLU-Pro | 78.8% | — |
| Confabulations | 14.2% | — |
| Vectara Hallucination Rate | 5.8% | — |
| GPQA (HELM) | 65% | — |
Multimodal Not comparable
Grok 3: —, Qwen3.5 397B-A17B: 40.0 (#42)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Vision | — | 1262 |
Multilingual Qwen3.5 397B-A17B leads
Grok 3: 52.5 (#81), Qwen3.5 397B-A17B: 54.0 (#56)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Non-English | 1411 | 1431 |
| LMArena Chinese | 1448 | 1501 |
| LMArena French | 1460 | 1461 |
| LMArena German | 1431 | 1447 |
| LMArena Japanese | 1389 | 1427 |
| LMArena Korean | 1374 | 1382 |
| LMArena Russian | 1416 | 1429 |
| LMArena Spanish | 1417 | 1443 |
Instruction Following Too close to call
Grok 3: 75.2 (#69), Qwen3.5 397B-A17B: 75.2 (#73)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Instruction Following | 1409 | 1424 |
| IFEval | 88.4% | — |
Long Context Qwen3.5 397B-A17B leads
Grok 3: 38.7 (#187), Qwen3.5 397B-A17B: 44.1 (#71)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Longer Query | 1439 | 1442 |
| Fiction.LiveBench | 58.3% | — |
Writing & Preference Qwen3.5 397B-A17B leads
Grok 3: 56.0 (#136), Qwen3.5 397B-A17B: 62.5 (#74)
| Benchmark | Grok 3 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Text | 1426 | 1438 |
| LMArena Creative Writing | 1414 | 1401 |
| EQ-Bench Creative Writing | 1186 | 1478 |
| LMArena Multi-Turn | 1424 | 1447 |
| Short-Story Creative Writing | 76.4% | — |
| WildBench | 84.9% | — |
Frequently asked questions
Is Grok 3 better than Qwen3.5 397B-A17B?
Qwen3.5 397B-A17B is the stronger model overall, scoring 45.7 to 39.6 on the Noometry Index.
Is Grok 3 or Qwen3.5 397B-A17B better for coding?
They score almost the same on coding (41.5 vs 41.7); test both on your own repository before choosing.
How many benchmarks do Grok 3 and Qwen3.5 397B-A17B share?
22 benchmarks have published results for both models. Grok 3 has 40 scored results on Noometry and Qwen3.5 397B-A17B has 36.