# DeepSeek-V3.1 vs Step 3.7 Flash

> DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 37.3 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v3-1-vs-step-3-7-flash
- Last updated: 2026-10-10
- Shared benchmarks: 0

## Summary

- The widest gap is in reasoning, where DeepSeek-V3.1 leads 27.9 to 21.6.
- Both cost about the same: $0.25 input and $0.95 output per million tokens.
- Step 3.7 Flash accepts more context: 256K tokens versus 164K.

## Snapshot

| | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| Provider | DeepSeek | StepFun |
| Noometry Index | 42.8 | 37.3 |
| Rank | 108 | 207 |
| Context | 164K | 256K |
| Input $/M | $0.25 | $0.18 |
| Output $/M | $0.95 | $1.11 |
| Weights | Open | Open |

## Coding

- DeepSeek-V3.1: 40.3 (#144)
- Step 3.7 Flash: 40.0 (#150)

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| SciCode | — | 40% |
| WeirdML | 38.4% | — |
| LMArena Coding | 1417 | — |
| ALE-Bench | — | 694.12 |

## Reasoning

- DeepSeek-V3.1: 27.9 (#110)
- Step 3.7 Flash: 21.6 (#219)

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| SimpleBench | 40% | — |
| Kagi LLM Benchmark | 53.2% | — |
| NYT Connections (extended) | — | 39.7% |
| CritPt | — | 2.3% |
| LMArena Hard Prompts | 1417 | — |
| DTBench | 82.7% | — |
| LMCA | 24.3% | — |
| Epoch Capabilities Index | 139.92 | — |
| ForecastBench | 58 | — |

## Math

- DeepSeek-V3.1: 38.9 (#122)
- Step 3.7 Flash: 42.9 (#82)

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| MathArena Final-Answer Competitions | — | 68.5% |
| LMArena Math | 1420 | — |

## Knowledge

- DeepSeek-V3.1: 43.7 (#90)
- Step 3.7 Flash: —

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| Vectara Hallucination Rate | 5.5% | — |
| LMArena Expert | 1405 | — |

## Multilingual

- DeepSeek-V3.1: 51.6 (#106)
- Step 3.7 Flash: —

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| LMArena Non-English | 1400 | — |
| LMArena Chinese | 1469 | — |
| LMArena French | 1447 | — |
| LMArena German | 1411 | — |
| LMArena Japanese | 1378 | — |
| LMArena Korean | 1337 | — |
| LMArena Russian | 1405 | — |
| LMArena Spanish | 1431 | — |

## Instruction Following

- DeepSeek-V3.1: 73.9 (#110)
- Step 3.7 Flash: —

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| LMArena Instruction Following | 1400 | — |

## Long Context

- DeepSeek-V3.1: 36.3 (#232)
- Step 3.7 Flash: —

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| Fiction.LiveBench | 52.8% | — |
| LMArena Longer Query | 1422 | — |

## Writing & Preference

- DeepSeek-V3.1: 60.3 (#98)
- Step 3.7 Flash: —

| Benchmark | DeepSeek-V3.1 | Step 3.7 Flash |
|---|---|---|
| LMArena Text | 1420 | — |
| LMArena Creative Writing | 1401 | — |
| EQ-Bench Creative Writing | 1436 | — |
| LMArena Multi-Turn | 1408 | — |

## FAQ

### Is DeepSeek-V3.1 better than Step 3.7 Flash?

DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 37.3 on the Noometry Index.

### Which is cheaper, DeepSeek-V3.1 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; DeepSeek-V3.1 lists at $0.25 and $0.95.

### Is DeepSeek-V3.1 or Step 3.7 Flash better for coding?

They score almost the same on coding (40.3 vs 40.0); test both on your own repository before choosing.

### Which has the bigger context window?

Step 3.7 Flash does, with 256K tokens against 164K.

### How many benchmarks do DeepSeek-V3.1 and Step 3.7 Flash share?

0 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Step 3.7 Flash has 5.