Step-3.5-Flash
Step 3.5 Flash (visit website) is our most capable open-source foundation model, engineered to deliver frontier reasoning and agentic capabilities with exceptional efficiency. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token. This "intelligence density" allows it to rival the reasoning depth of top-tier proprietary models, while maintaining th...
Params
199.4 B
Context
262,144
Downloads 30d
264 K
Likes
844
Download history
daily snapshots · 21 days269 K195 K
Oct 4Oct 11Oct 18Oct 24
Can you run it?
Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.
| File | Quant | Size | Est. VRAM | Verdict on RTX 4090 · 24 GB |
|---|---|---|---|---|
| model.safetensors | bf16 | 398.8 GB | 469.1 GB | ❌ Won’t fit |
| model.safetensors (bf16, full) | bf16 + 256K ctx | 398.8 GB | 1,396.2 GB | ❌ Won’t fit |
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.
Run it
copy-paste, exact tags checked against the Hub$ curl -s https://aimodelscomparison.com/api/v1/models/step-3-5-flash
{
"hf_id": "stepfun-ai/Step-3.5-Flash",
"params_b": 199.38,
"context_length": 262144,
"license": { "id": "apache-2.0", "commercial": "yes" },
"downloads_30d": 264309,
"vram_estimates": [
{ "quant": "bf16", "gb": 469.1 }
],
"updated_at": "2026-10-04T01:00:14Z"
}
Use via API
via OpenRouter · USD per 1M tokens · synced 2026-10-24$0.10
input / 1M
$0.30
output / 1M
—
cached input / 1M
$0.40
1M in + 1M out
- Context window
- 262,144 tokens
- Max output
- 65,536 tokens
- Providers
- 1
- Cheapest provider
- SiliconFlow · $0.10 / $0.30
- Best throughput
- —
- AA Intelligence / Coding / Agentic
- 17 / — / —
✓ Tool calling
✓ Reasoning
✗ Structured outputs
✗ Image input
✗ File input
✗ Audio input
✗ Prompt caching
Specifications
- Architecture
- Step3p5ForCausalLM
- Parameters
- 199.4 B
- Tensor type
- BF16
- Context length
- 262,144
- Vocabulary
- 128,896
- Layers / heads
- 45 / 64
- Licence
- apache-2.0
- First seen on the Hub
- 2026-02-01
- Base model
- Step-3.5-Flash
- Training datasets
- undisclosed
- Added to our catalog
- 2026-10-04
Family
Base model and the most-downloaded derivatives in the catalog.
Compare with any text-generation model