Model comparison
Inkling vs Qwen3.5 397B-A17B
Qwen3.5 397B-A17B is the stronger model overall, scoring 45.7 to 44.1 on the Noometry Index.
Last verified . 28 shared benchmarks.
Alibaba (Qwen)Summary
- They share 28 benchmarks with published results for both. Inkling scores higher in 5 categories and Qwen3.5 397B-A17B in 4 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3.5 397B-A17B leads 44.9 to 31.2.
- The biggest single-benchmark swing is τ²-bench Banking: 25% for Inkling and 9.8% for Qwen3.5 397B-A17B.
- Qwen3.5 397B-A17B is cheaper at $0.60 / $3.60 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- Qwen3.5 397B-A17B accepts more context: 262K tokens versus 66K.
Side by side
| Inkling | Qwen3.5 397B-A17B | |
|---|---|---|
| Provider | Thinking Machines Lab | Alibaba (Qwen) |
| Noometry Index | 44.1 | 45.7 |
| Released | 2026-07-15 | 2026-02-01 |
| Weights | Open | Open |
| Context window | 66K | 262K |
| Max output | 66K | 66K |
| Input $ / M tokens | $1.87 | $0.60 |
| Output $ / M tokens | $4.68 | $3.60 |
| Results tracked | 41 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.5 397B-A17B leads
Inkling: 34.1 (#229), Qwen3.5 397B-A17B: 41.7 (#108)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena WebDev | 1412 | 1400 |
| LMArena Coding | 1463 | 1464 |
| FrontierCode | 14% | — |
| FrontierSWE | 4.1% | — |
| SciCode | 47% | — |
| WeirdML | 32.3% | — |
| ALE-Bench | 946 | — |
Agentic & Tool Use Qwen3.5 397B-A17B leads
Inkling: 29.5 (#76), Qwen3.5 397B-A17B: 33.2 (#51)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| APEX-Agents | 33.8% | 24.9% |
| τ²-bench Banking | 25% | 9.8% |
| τ²-bench Airline | — | 81.5% |
| τ²-bench Retail | — | 84.4% |
| τ²-bench Telecom | — | 97.8% |
Reasoning Inkling leads
Inkling: 40.6 (#48), Qwen3.5 397B-A17B: 33.6 (#62)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| Chess Puzzles | 21% | 13% |
| LMArena Hard Prompts | 1452 | 1448 |
| DTBench | 87.5% | 87.5% |
| LMCA | 37.6% | 37.9% |
| Epoch Capabilities Index | 148.6 | 146.64 |
| ARC-AGI-2 | 36.5% | — |
| SimpleBench | 50% | — |
| Kagi LLM Benchmark | — | 73.7% |
| NYT Connections (extended) | — | 58.9% |
| ARC-AGI-1 | 79.5% | — |
| CritPt | 5.4% | — |
| Thematic Generalization | — | 65.1% |
| Mystery Game Puzzles | — | 18% |
Math Qwen3.5 397B-A17B leads
Inkling: 31.2 (#198), Qwen3.5 397B-A17B: 44.9 (#64)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| FrontierMath (Tiers 1-3) | 33.3% | 31.2% |
| OTIS Mock AIME 2024-2025 | 88.9% | 88.9% |
| LMArena Math | 1482 | 1455 |
| FrontierMath Tier 4 | 4.9% | — |
| ProofBench | 0% | — |
Knowledge Inkling leads
Inkling: 55.2 (#44), Qwen3.5 397B-A17B: 53.3 (#53)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| GPQA Diamond | 88.3% | 86.4% |
| LMArena Expert | 1468 | 1461 |
| SimpleQA Verified | 40.3% | — |
Multimodal Not comparable
Inkling: —, Qwen3.5 397B-A17B: 40.0 (#42)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Vision | — | 1262 |
Multilingual Too close to call
Inkling: 54.2 (#49), Qwen3.5 397B-A17B: 54.0 (#56)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Non-English | 1434 | 1431 |
| LMArena Chinese | 1495 | 1501 |
| LMArena French | 1453 | 1461 |
| LMArena German | 1456 | 1447 |
| LMArena Japanese | 1428 | 1427 |
| LMArena Korean | 1401 | 1382 |
| LMArena Russian | 1430 | 1429 |
| LMArena Spanish | 1442 | 1443 |
Instruction Following Too close to call
Inkling: 75.3 (#65), Qwen3.5 397B-A17B: 75.2 (#73)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Instruction Following | 1427 | 1424 |
Long Context Too close to call
Inkling: 43.9 (#79), Qwen3.5 397B-A17B: 44.1 (#71)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Longer Query | 1436 | 1442 |
Writing & Preference Inkling leads
Inkling: 65.5 (#48), Qwen3.5 397B-A17B: 62.5 (#74)
| Benchmark | Inkling | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Text | 1443 | 1438 |
| LMArena Creative Writing | 1390 | 1401 |
| EQ-Bench Creative Writing | 1611 | 1478 |
| LMArena Multi-Turn | 1436 | 1447 |
| EQ-Bench 4 | 1226 | — |
Frequently asked questions
Is Inkling better than Qwen3.5 397B-A17B?
Qwen3.5 397B-A17B is the stronger model overall, scoring 45.7 to 44.1 on the Noometry Index.
Which is cheaper, Inkling 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; Inkling lists at $1.87 and $4.68.
Is Inkling or Qwen3.5 397B-A17B better for coding?
Qwen3.5 397B-A17B scores higher on coding benchmarks: 41.7 versus 34.1 in the Noometry coding category.
Which has the bigger context window?
Qwen3.5 397B-A17B does, with 262K tokens against 66K.
How many benchmarks do Inkling and Qwen3.5 397B-A17B share?
28 benchmarks have published results for both models. Inkling has 41 scored results on Noometry and Qwen3.5 397B-A17B has 36.