Model comparison
Qwen3.5 397B-A17B vs Step 3.7 Flash
Qwen3.5 397B-A17B is the stronger model overall, scoring 45.7 to 36.5 on the Noometry Index. Step 3.7 Flash costs 3.2× less per token, which makes it the better buy when Qwen3.5 397B-A17B's lead doesn't matter for your workload.
Last verified . 1 shared benchmarks.
Alibaba (Qwen)
StepFunSummary
- They share 1 benchmark with published results for both. Qwen3.5 397B-A17B scores higher in 3 categories and Step 3.7 Flash in 0 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.5 397B-A17B leads 33.6 to 20.9.
- The biggest single-benchmark swing is NYT Connections (extended): 58.9% for Qwen3.5 397B-A17B and 39.7% for Step 3.7 Flash.
- Step 3.7 Flash is cheaper at $0.18 / $1.11 per million input/output tokens, against $0.60 / $3.60 for Qwen3.5 397B-A17B.
- Qwen3.5 397B-A17B accepts more context: 262K tokens versus 256K.
Side by side
| Qwen3.5 397B-A17B | Step 3.7 Flash | |
|---|---|---|
| Provider | Alibaba (Qwen) | StepFun |
| Noometry Index | 45.7 | 36.5 |
| Released | 2026-02-01 | 2026-05-29 |
| Weights | Open | Open |
| Context window | 262K | 256K |
| Max output | 66K | 256K |
| Input $ / M tokens | $0.60 | $0.18 |
| Output $ / M tokens | $3.60 | $1.11 |
| Results tracked | 36 | 5 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.5 397B-A17B leads
Qwen3.5 397B-A17B: 41.7 (#108), Step 3.7 Flash: 39.4 (#147)
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| LMArena WebDev | 1400 | — |
| SciCode | — | 40% |
| LMArena Coding | 1464 | — |
| ALE-Bench | — | 694.12 |
Agentic & Tool Use Not comparable
Qwen3.5 397B-A17B: 33.2 (#51), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| 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
Qwen3.5 397B-A17B: 33.6 (#62), Step 3.7 Flash: 20.9 (#211)
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| NYT Connections (extended) | 58.9% | 39.7% |
| Kagi LLM Benchmark | 73.7% | — |
| CritPt | — | 2.3% |
| Chess Puzzles | 13% | — |
| Thematic Generalization | 65.1% | — |
| LMArena Hard Prompts | 1448 | — |
| Mystery Game Puzzles | 18% | — |
| DTBench | 87.5% | — |
| LMCA | 37.9% | — |
| Epoch Capabilities Index | 146.64 | — |
Math Qwen3.5 397B-A17B leads
Qwen3.5 397B-A17B: 44.9 (#64), Step 3.7 Flash: 40.9 (#74)
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| FrontierMath (Tiers 1-3) | 31.2% | — |
| MathArena Final-Answer Competitions | — | 68.5% |
| OTIS Mock AIME 2024-2025 | 88.9% | — |
| LMArena Math | 1455 | — |
Knowledge Not comparable
Qwen3.5 397B-A17B: 53.3 (#53), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| GPQA Diamond | 86.4% | — |
| LMArena Expert | 1461 | — |
Multimodal Not comparable
Qwen3.5 397B-A17B: 40.0 (#42), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| LMArena Vision | 1262 | — |
Multilingual Not comparable
Qwen3.5 397B-A17B: 54.0 (#56), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| LMArena Non-English | 1431 | — |
| LMArena Chinese | 1501 | — |
| LMArena French | 1461 | — |
| LMArena German | 1447 | — |
| LMArena Japanese | 1427 | — |
| LMArena Korean | 1382 | — |
| LMArena Russian | 1429 | — |
| LMArena Spanish | 1443 | — |
Instruction Following Not comparable
Qwen3.5 397B-A17B: 75.2 (#73), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| LMArena Instruction Following | 1424 | — |
Long Context Not comparable
Qwen3.5 397B-A17B: 44.1 (#71), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| LMArena Longer Query | 1442 | — |
Writing & Preference Not comparable
Qwen3.5 397B-A17B: 62.5 (#74), Step 3.7 Flash: —
| Benchmark | Qwen3.5 397B-A17B | Step 3.7 Flash |
|---|---|---|
| LMArena Text | 1438 | — |
| LMArena Creative Writing | 1401 | — |
| EQ-Bench Creative Writing | 1478 | — |
| LMArena Multi-Turn | 1447 | — |
Frequently asked questions
Is Qwen3.5 397B-A17B better than Step 3.7 Flash?
Qwen3.5 397B-A17B is the stronger model overall, scoring 45.7 to 36.5 on the Noometry Index. Step 3.7 Flash costs 3.2× 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, Qwen3.5 397B-A17B or Step 3.7 Flash?
Step 3.7 Flash is cheaper. It lists at $0.18 per million input tokens and $1.11 per million output tokens; Qwen3.5 397B-A17B lists at $0.60 and $3.60.
Is Qwen3.5 397B-A17B or Step 3.7 Flash better for coding?
Qwen3.5 397B-A17B scores higher on coding benchmarks: 41.7 versus 39.4 in the Noometry coding category.
Which has the bigger context window?
Qwen3.5 397B-A17B does, with 262K tokens against 256K.
How many benchmarks do Qwen3.5 397B-A17B and Step 3.7 Flash share?
1 benchmark has published results for both models. Qwen3.5 397B-A17B has 36 scored results on Noometry and Step 3.7 Flash has 5.