z-lab / text-generation updated 1 month ago

Qwen3.6-35B-A3B-DFlash

Name: Qwen3.6-35B-A3B-DFlash
Author: z-lab

This DFlash draft model is a joint retrain from Z-Lab and Modal, trained with 40k sequence length and sliding-window attention for improved long-context performance. It is mirrored across the following Hugging Face repositories:

Params

390 M

Context

262,144

Downloads 30d

228 K

Likes

285

Commercial use: allowed apache-2.0 Not gated SAFETENSORS View on Hugging Face ↗

Download history

daily snapshots · 10 days

233 K222 K

Jul 28Jul 31Aug 3Aug 6

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	0.8 GB	1.4 GB	✅ Runs comfortably
model.safetensors (bf16, full)	bf16 + 256K ctx	0.8 GB	3.2 GB	✅ Runs comfortably

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 · api/v1

$ curl -s https://aimodelscomparison.com/api/v1/models/qwen3-6-35b-a3b-dflash
{
  "hf_id": "z-lab/Qwen3.6-35B-A3B-DFlash",
  "params_b": 0.39,
  "context_length": 262144,
  "license": { "id": "apache-2.0", "commercial": "yes" },
  "downloads_30d": 227605,
  "vram_estimates": [
    { "quant": "bf16", "gb": 1.4 }
  ],
  "updated_at": "2026-07-28T18:07:34Z"
}

est. VRAM —on RTX 4090 · 24 GBJSON API →

Specifications

Architecture: DFlashDraftModel
Parameters: 390 M
Tensor type: BF16
Context length: 262,144
Vocabulary: 248,320
Layers / heads: 6 / 32
Licence: apache-2.0
First seen on the Hub: 2026-04-17
Base model: Qwen3.6-35B-A3B
Training datasets: undisclosed
Added to our catalog: 2026-07-28

Family

Base model and the most-downloaded derivatives in the catalog.

Base Qwen3.6-35B-A3B Qwen 5.9 M

Compare with

Qwen3-0.6B 29.7 M Qwen2.5-1.5B-Instruct 14.0 M gpt2 13.7 M Llama-3.2-1B-Instruct 10.4 M Qwen2.5-0.5B-Instruct 6.6 M Qwen2.5-0.5B 2.6 M

Appears in

Best open coding models Commercial-use text models Models that fit in 12 GB