BestLLMfor EN Your hardware. Your LLM. Your call.
APIOpen data Find my LLM
Model fiche

Step 3.5 Flash

By StepFun · China

Updated 2026-07-13

chat general moe
Parameters
196B
License
Apache 2.0
Context
250k
VRAM (Q4)
118 GB
Released
February 2026

Overview

StepFun's 196B MoE with 11B active parameters delivers 100 tokens/sec at 128K context. Ranks #3 by free-tier volume on OpenRouter under Apache 2.0.

When to pick this model

  • High-throughput chat backends
  • Long-context workloads needing fast inference
  • Apache-licensed commercial deployments
  • Cost-sensitive production at scale
  • Workloads where latency matters more than top quality

VRAM requirements by quantization

VRAM REQUIRED (GB)324880128256Q4_K_M118 GBQ5_K_M141 GBQ8_0210 GBFP16392 GB
QuantizationVRAM required
Q4_K_M (recommended)118 GB
Q5_K_M141 GB
Q8_0210 GB
FP16 (no quantization)392 GB

VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.

In practice, Step 3.5 Flash is server-class even at Q4_K_M (118 GB). Stepping up to Q8_0 nearly doubles the footprint to 210 GB, and unquantized FP16 weights take 392 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.

Without a GPU, Step 3.5 Flash needs roughly 140 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 2 tokens/sec on entry-level GPUs, on the order of 10 tokens/sec on a mid-range card, and up to 25 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.

What hardware do you need

The table below matches Step 3.5 Flash to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.

GPU memoryExample cardsBest fit for Step 3.5 Flash
8 GBRTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GBDoes not fit — needs 118 GB at Q4_K_M
12 GBRTX 5070, RTX 5070 Ti Laptop, RTX 4080 LaptopDoes not fit — needs 118 GB at Q4_K_M
16 GBRTX 5080, RTX 4080 Super, Radeon RX 9070 XTDoes not fit — needs 118 GB at Q4_K_M
24 GBRTX 4090, Radeon RX 7900 XTX, RTX 5090 LaptopDoes not fit — needs 118 GB at Q4_K_M
32 GBRTX 5090Does not fit — needs 118 GB at Q4_K_M

Which GPU should you buy to run Step 3.5 Flash?

To run Step 3.5 Flash locally at Q4, you need ~118 GB of VRAM. The best value for this is a Apple Mac Studio (64+ GB unified memory).

Check Apple Mac Studio price on Amazon →

As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.

Strengths

  • 100 tokens/sec sustained at 128K context
  • 256K maximum context window
  • Only 11B active parameters
  • Apache 2.0 license

Limitations

  • 118GB+ in Q4 needs a multi-GPU server
  • Brand awareness still low outside Asia
  • Trails top open models on hardest benchmarks

Typical workloads

In our catalog grid, Step 3.5 Flash is filed under Fast MoE, Long Context — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: everyday chat, drafting and summarization.

The 250k-token context window is large enough to hold entire codebases' worth of files or long reports in a single prompt, which is what makes local RAG and document analysis practical. The Apache 2.0 license is permissive, so shipping it inside a commercial product raises no special legal questions.

Architecture & training

Architecture: MoE 196B/11B active · 256k ctx

Training: StepFun. 100 tok/s at 128k ctx.

Verdict

A fast, permissively licensed MoE that punches well above its name recognition.

Quick start

# HuggingFace : stepfun-ai/step-3.5-flash

Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.

Similar models worth comparing

Frequently asked questions

How much VRAM does Step 3.5 Flash need?

At the recommended Q4_K_M quantization, Step 3.5 Flash needs about 118 GB of VRAM. Q8_0 takes 210 GB, and unquantized FP16 weights take 392 GB.

Can Step 3.5 Flash run without a GPU?

Yes — with roughly 140 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.

What context window does Step 3.5 Flash support?

Step 3.5 Flash supports a 250k-token context window (256,000 tokens).

Can I use Step 3.5 Flash commercially?

Yes. Step 3.5 Flash is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.

How fast is Step 3.5 Flash on consumer hardware?

Our compatibility engine estimates on the order of 10 tokens/sec on a mid-range GPU and up to 25 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.

Which quantization of Step 3.5 Flash should I download first?

Start with Q4_K_M (118 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It does not fit a single 24 GB consumer card — plan for multi-GPU or server hardware.

Tools

Is Step 3.5 Flash the right pick for you?

Compute self-hosted ROI → Back to catalog