Step-3.5-Flash
StepFun- Arena score
- 1402
- Among open models
- #47of 148
- Among all on Arena
- #154
- Votes
- 59,749
Measured on Step-3.5-Flash, the served version, not a local build. LMArena, CC BY 4.0, published Nov 11, 2026.
Score by category
CategoryScoreOpen rankArena rankVotes
- Chinese1446#41#1353.6K
- Coding1435#40#13215.1K
- Software & IT1434#43#13622.6K
- French1421#39#1312K
- Expert1421#45#1385.2K
- Hard prompts, English1420#47#14716.1K
- Applied math1419#44#1393.1K
- Spanish1418#36#1202K
- English1417#48#14725K
- Science1414#47#1569.7K
- Hard prompts1410#48#15337.2K
- Medicine & health1408#44#1514.2K
- Legal & government1407#45#1534.6K
- Math1406#45#1533.4K
- Multi-turn1404#46#15110.3K
- German1403#36#1261.1K
- Long prompts1402#44#14922.5K
- Business & finance1394#49#15411.7K
- Polish1388#38#1391.4K
- Russian1385#47#1617.4K
- Non-English1385#47#15834.7K
- Instructions1385#46#15519.1K
- Excluding ties1384#48#15543.5K
- Writing & literature1371#43#16014.6K
- Media & sports1359#46#16011.9K
- Creative writing1357#43#1629.7K
- Japanese1353#32#128641
- Korean1350#40#1421.1K
About as strong, smaller
Compare
gemma-4-26B-A4B-itGoogle143425.8B

Qwen3.8-27BQwen143727.8B

Qwen3.5-27BQwen140927.8B

Muse-Glimmer-30Bmeta-models138929.8B
Discussion
Built on it
Ways to run it
- GGUF · llama.cpp, Ollama, LM Studioggml-org/Step-3.5-Flash-GGUF13 builds
- MLX · Apple silicon: mlx-lm, LM Studiomlx-community/Step-3.5-Flash-4bit4 builds
Most downloaded
- Step-3.5-Flash-GGUFggml-org · Quantization20.9K downloads
- stepfun-ai_Step-3.5-Flash-GGUFbartowski · Quantization1.3K downloads
- Step-3.5-Flash-APEX-GGUFmudler · Quantization978 downloads
- Step-3.5-Flash-i1-GGUFmradermacher · Quantization745 downloads
- Step-3.5-Flash-4bitmlx-community · Quantization502 downloads
BenchClub runs
Not runnable yet
Runs open once a Step-3.5-Flash build is registered.
Other 80–300B models
See more
DeepSeek-V4-FlashDeepSeek1.2M291B

MiniMax-M2.5MiniMax1M229B
Inkling-SmallThinking Machines1M266B
Hy3Tencent987.2K299B