Model comparison
Grok 4.6 vs Qwen3.5 397B-A17B
Grok 4.6 is the stronger model overall, scoring 56.9 to 46.0 on the Noometry Index. Qwen3.5 397B-A17B costs 2.2× less per token, which makes it the better buy when Grok 4.6's lead doesn't matter for your workload.
Last verified . 29 shared benchmarks.
Alibaba (Qwen)Summary
- They share 29 benchmarks with published results for both. Grok 4.6 scores higher in 8 categories and Qwen3.5 397B-A17B in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Grok 4.6 leads 61.4 to 34.5.
- The biggest single-benchmark swing is APEX-Agents: 65.3% for Grok 4.6 and 24.9% for Qwen3.5 397B-A17B.
- Qwen3.5 397B-A17B is cheaper at $0.60 / $3.60 per million input/output tokens, against $2 / $6 for Grok 4.6.
- Grok 4.6 accepts more context: 500K tokens versus 262K.
- Qwen3.5 397B-A17B has downloadable open weights; the other is API-only.
Side by side
| Grok 4.6 | Qwen3.5 397B-A17B | |
|---|---|---|
| Provider | xAI | Alibaba (Qwen) |
| Noometry Index | 56.9 | 46.0 |
| Released | 2026-08-12 | 2026-02-01 |
| Weights | Proprietary | Open |
| Context window | 500K | 262K |
| Max output | 500K | 66K |
| Input $ / M tokens | $2 | $0.60 |
| Output $ / M tokens | $6 | $3.60 |
| Results tracked | 49 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.6 leads
Grok 4.6: 58.5 (#16), Qwen3.5 397B-A17B: 42.0 (#114)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena WebDev | 1617 | 1400 |
| LMArena Coding | 1465 | 1465 |
| DeepSWE | 67.5% | — |
| FrontierCode | 48% | — |
| CursorBench | 41.4% | — |
| FrontierSWE | 25.3% | — |
| SciCode | 56.5% | — |
| WeirdML | 67.3% | — |
| ALE-Bench | 1,508 | — |
Agentic & Tool Use Grok 4.6 leads
Grok 4.6: 39.4 (#27), Qwen3.5 397B-A17B: 33.3 (#53)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| APEX-Agents | 65.3% | 24.9% |
| τ²-bench Airline | — | 81.5% |
| τ²-bench Banking | — | 9.8% |
| τ²-bench Retail | — | 84.4% |
| τ²-bench Telecom | — | 97.8% |
| GDP.pdf | 17.2% | — |
| Vending-Bench 2 | 9,047 | — |
Reasoning Grok 4.6 leads
Grok 4.6: 61.4 (#20), Qwen3.5 397B-A17B: 34.5 (#70)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| NYT Connections (extended) | 80% | 58.9% |
| Chess Puzzles | 40% | 13% |
| LMArena Hard Prompts | 1447 | 1448 |
| Mystery Game Puzzles | 34% | 18% |
| DTBench | 97.3% | 87.5% |
| LMCA | 48.5% | 37.9% |
| Epoch Capabilities Index | 156.44 | 146.65 |
| ARC-AGI-2 | 67.1% | — |
| SimpleBench | 75.9% | — |
| Kagi LLM Benchmark | — | 73.7% |
| ARC-AGI-1 | 87.5% | — |
| CritPt | 19.7% | — |
| Thematic Generalization | — | 65.1% |
| EBR-Bench | 30.5% | — |
Math Grok 4.6 leads
Grok 4.6: 67.0 (#24), Qwen3.5 397B-A17B: 46.1 (#73)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| FrontierMath (Tiers 1-3) | 66% | 31.2% |
| OTIS Mock AIME 2024-2025 | 99.2% | 88.9% |
| LMArena Math | 1423 | 1454 |
| FrontierMath Tier 4 | 31.7% | — |
| ProofBench | 51% | — |
Knowledge Grok 4.6 leads
Grok 4.6: 63.3 (#20), Qwen3.5 397B-A17B: 53.3 (#58)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| GPQA Diamond | 94% | 86.4% |
| LMArena Expert | 1467 | 1462 |
| SimpleQA Verified | 49.3% | — |
Multimodal Grok 4.6 leads
Grok 4.6: 43.6 (#23), Qwen3.5 397B-A17B: 40.7 (#44)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Vision | 1263 | 1263 |
| Blueprint-Bench 2 | 33.2% | — |
| Furniture Assembly | 40% | — |
| LMArena Document | 1452 | — |
Multilingual Too close to call
Grok 4.6: 53.0 (#74), Qwen3.5 397B-A17B: 53.7 (#59)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Non-English | 1420 | 1430 |
| LMArena Chinese | 1480 | 1500 |
| LMArena French | 1461 | 1461 |
| LMArena German | 1431 | 1447 |
| LMArena Japanese | 1376 | 1426 |
| LMArena Korean | 1397 | 1384 |
| LMArena Russian | 1422 | 1429 |
| LMArena Spanish | 1404 | 1441 |
Instruction Following Too close to call
Grok 4.6: 75.4 (#63), Qwen3.5 397B-A17B: 75.0 (#77)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Instruction Following | 1431 | 1424 |
Long Context Too close to call
Grok 4.6: 44.5 (#66), Qwen3.5 397B-A17B: 44.1 (#74)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Longer Query | 1454 | 1442 |
Writing & Preference Too close to call
Grok 4.6: 62.3 (#80), Qwen3.5 397B-A17B: 62.3 (#79)
| Benchmark | Grok 4.6 | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Text | 1428 | 1438 |
| LMArena Creative Writing | 1428 | 1401 |
| LMArena Multi-Turn | 1425 | 1446 |
| EQ-Bench Creative Writing | — | 1478 |
Frequently asked questions
Is Grok 4.6 better than Qwen3.5 397B-A17B?
Grok 4.6 is the stronger model overall, scoring 56.9 to 46.0 on the Noometry Index. Qwen3.5 397B-A17B costs 2.2× less per token, which makes it the better buy when Grok 4.6's lead doesn't matter for your workload.
Which is cheaper, Grok 4.6 or Qwen3.5 397B-A17B?
Qwen3.5 397B-A17B is cheaper. It lists at $0.60 per million input tokens and $3.60 per million output tokens; Grok 4.6 lists at $2 and $6.
Is Grok 4.6 or Qwen3.5 397B-A17B better for coding?
Grok 4.6 scores higher on coding benchmarks: 58.5 versus 42.0 in the Noometry coding category.
Which has the bigger context window?
Grok 4.6 does, with 500K tokens against 262K.
How many benchmarks do Grok 4.6 and Qwen3.5 397B-A17B share?
29 benchmarks have published results for both models. Grok 4.6 has 49 scored results on Noometry and Qwen3.5 397B-A17B has 36.