sakamakismile / text-generation Open weights updated 2 months ago

Qwen3.8-27B-MTP-NVFP4

Name: Qwen3.8-27B-MTP-NVFP4
Author: sakamakismile

Qwen/Qwen3.8-27B (dense 27.8B hybrid DeltaNet + full-attention, Qwen3-VL vision, native MTP head) quantized to NVFP4 (W4A4, group 16) with llm-compressor. 55.6 GB bf16 → 20.6 GB.

Params

27.4 B

Context

—

Downloads 30d

313 K

Likes

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

Download history

daily snapshots · 50 days

▲ 141 K in the last 30 days (31.1%)

455 K200 K

Sep 22Oct 8Oct 25Nov 10

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	u8	20.6 GB	27.2 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 · api/v1

$ curl -s https://aimodelscomparison.com/api/v1/models/qwen3-8-27b-mtp-nvfp4
{
  "hf_id": "sakamakismile/Qwen3.8-27B-MTP-NVFP4",
  "params_b": 27.36,
  "context_length": null,
  "license": { "id": "apache-2.0", "commercial": "yes" },
  "downloads_30d": 313117,
  "vram_estimates": [
    { "quant": "u8", "gb": 27.2 }
  ],
  "updated_at": "2026-09-22T01:00:19Z"
}

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

Specifications

Architecture: Qwen3_5ForConditionalGeneration
Parameters: 27.4 B
Tensor type: U8
Licence: apache-2.0
First seen on the Hub: 2026-08-14
Base model: Qwen3.8-27B
Training datasets: undisclosed
Added to our catalog: 2026-09-22

Family

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

Base Qwen3.8-27B Qwen 7.5 M

Compare with any text-generation model

Popular comparisons

Qwen3-14B 12.9 M gpt-oss-20b 5.8 M dolphin-2.9.1-yi-1.5-34b 4.8 M Qwen3.6-35B-A3B-NVFP4 4.0 M OTel-2.0-Decision-26B-A4B-4bit 3.2 M Qwen3-32B 3.0 M