Model reference · open weights
Qwen3.8-DFlash2 is an open-weight language model from incoai, listed in the AxForge catalogue. AxForge can bring it up on EU-owned hardware for you on request — with the licence handled where one is required.
About
Qwen3.8-27B-DFlash2 Blog | GitHub This repository contains the DFlash 2 draft model for Qwen/Qwen3.8-27B. It is not a standalone language model: it runs inside a speculative decoding server and drafts tokens for the target model to verify. The checkpoint is also mirrored at z-lab/Qwen3.8-27B-DFlash2. DFlash 2 is a block-diffusion drafter for speculative decoding. It predicts a whole block of tokens in a single pass and keeps the top candidates at every position. A lightweight selector then traces one coherent path through them. Two-tap dynamic convolutions in the backbone keep the draft from decaying toward the end of the block. Decoding is lossless: greedy output matches the target model exactly, and sampling preserves its distribution. Quick Start Serve with SGLang: Or with vLLM: See the blog post for other engines and more details. Evaluation - Runtime: SGLang on one NVIDIA H200, with FlashAttention 3 for target and draft attention - Speculation block size: 8 (7 draft tokens per verification step) - Sampling: Qwen3.8's officially recommended parameters (temperature 1.0, top-p 0.95, top-k 20), with xhigh reasoning effort - Maximum new tokens: 4096 - Prompts: benchmark formatting from z-lab/dflash We compare autoregressive decoding, Qwen3.8's built-in seven-token MTP, a community DSpark drafter (RadixArk/Qwen3.8-27B-DSpark), and DFlash 2. All speculative methods propose seven draft tokens per verification step. Acceptance Length Acceptance length is the per-request mean of completion tokens divided by verification steps. Higher is better. Throughput Throughput is total output tokens divided by end-to-end wall time. Each cell shows output tok/s (speedup vs. autoregressive). Concurrency 1 Concurrency 8 Concurrency 32 Citation If you find DFlash 2 useful, please cite: Please also cite the original DFlash paper:
Summarised from the published model card. Read the full card on the HuggingFace links below.
Specifications
| Maker | incoai |
|---|---|
| Type | Language models |
| Parameters (lead) | 1.9B |
| Context | 256k tokens |
| Variants | 2 |
| Runs with | transformers |
| Based on | Qwen/Qwen3.8-27B |
| Released | 2026-08-18 |
| Popularity | 203k downloads / month |
| Likes | 202 |
| Licence | Open weights |
How it works
Variants
Open weights ship in several sizes and precisions. One page, all the variants — pick the one that fits your GPU. VRAM figures are estimates from model size.
Using it via the API
Once AxForge deploys qwen3-8-dflash2 for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (qwen3-8-dflash2 below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/chat/completions \
-H "Authorization: Bearer $AXFORGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"qwen3-8-dflash2","messages":[{"role":"user","content":"Hello"}]}'
Details
Tags
Licence
Open weights under apache-2.0 — commercial use is permitted. Deploy it on AxForge EU hardware on request. Read the licence ↗