# DeepSeek-V2.5 (Sep 2024) vs Step 3.5 Flash

> Step 3.5 Flash is the stronger model overall, scoring 41.7 to 37.0 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v2-5-vs-step-3-5-flash
- Last updated: 2026-10-10
- Shared benchmarks: 17

## Summary

- They share 17 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 1 category and Step 3.5 Flash in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in coding, where Step 3.5 Flash leads 41.7 to 31.3.

## Snapshot

| | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| Provider | DeepSeek | StepFun |
| Noometry Index | 37.0 | 41.7 |
| Rank | 197 | 116 |
| Context | — | 256K |
| Input $/M | — | $0.10 |
| Output $/M | — | $0.30 |
| Weights | Open | Open |

## Coding

- DeepSeek-V2.5 (Sep 2024): 31.3 (#274)
- Step 3.5 Flash: 41.7 (#106)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Coding | 1310 | 1436 |
| Aider Polyglot | 17.8% | — |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |

## Reasoning

- DeepSeek-V2.5 (Sep 2024): 24.5 (#145)
- Step 3.5 Flash: 21.2 (#202)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1411 |
| NYT Connections (extended) | — | 28.4% |

## Math

- DeepSeek-V2.5 (Sep 2024): 33.3 (#177)
- Step 3.5 Flash: 40.6 (#77)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Math | 1289 | 1408 |
| MathArena Final-Answer Competitions | — | 66.8% |

## Knowledge

- DeepSeek-V2.5 (Sep 2024): 34.2 (#190)
- Step 3.5 Flash: 39.0 (#127)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Expert | 1266 | 1422 |

## Multilingual

- DeepSeek-V2.5 (Sep 2024): 42.7 (#188)
- Step 3.5 Flash: 50.7 (#115)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Non-English | 1273 | 1386 |
| LMArena Chinese | 1317 | 1448 |
| LMArena French | 1288 | 1421 |
| LMArena German | 1258 | 1404 |
| LMArena Japanese | 1228 | 1353 |
| LMArena Korean | 1209 | 1350 |
| LMArena Russian | 1289 | 1385 |
| LMArena Spanish | 1248 | 1419 |

## Instruction Following

- DeepSeek-V2.5 (Sep 2024): 67.7 (#189)
- Step 3.5 Flash: 73.3 (#119)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Instruction Following | 1280 | 1385 |

## Long Context

- DeepSeek-V2.5 (Sep 2024): 39.5 (#169)
- Step 3.5 Flash: 42.8 (#112)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Longer Query | 1301 | 1402 |

## Writing & Preference

- DeepSeek-V2.5 (Sep 2024): 50.0 (#182)
- Step 3.5 Flash: 59.0 (#108)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Step 3.5 Flash |
|---|---|---|
| LMArena Text | 1294 | 1403 |
| LMArena Creative Writing | 1285 | 1357 |
| LMArena Multi-Turn | 1297 | 1405 |

## FAQ

### Is DeepSeek-V2.5 (Sep 2024) better than Step 3.5 Flash?

Step 3.5 Flash is the stronger model overall, scoring 41.7 to 37.0 on the Noometry Index.

### Is DeepSeek-V2.5 (Sep 2024) or Step 3.5 Flash better for coding?

Step 3.5 Flash scores higher on coding benchmarks: 41.7 versus 31.3 in the Noometry coding category.

### How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Step 3.5 Flash share?

17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Step 3.5 Flash has 19.