selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

Qwen logo

Qwen3.8-Flash-Next

Qwen·Instruct·released 24 Aug 2026

qwen-community-1.0180BA61.6BMoE256K ctx ≈108 GB
Your GPU
ollama run qwen3.8-flash-next
All 6 runners ↓

Qwen3.8-Flash-Next is an instruct-tuned vision-language model from Qwen. It uses a mixture-of-experts design with 180B total parameters and 61.6B active parameters. It accepts text and image inputs and returns text, with a context length of 262,144 tokens. It is released under the qwen-community-1.0 licence, which allows limited commercial use. No other sizes in a model family are listed for it.

How much VRAM does Qwen3.8-Flash-Next need?

At Q4_K_M Qwen3.8-Flash-Next needs about 108 GB with an 8K context (108 GB at Q4, 353 GB at full precision). Pick your GPU to see what fits and how fast it runs.

--vram
any
--gpus
--quant

Needs: ≈108 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.

Change quant, context or KV cache to compare
Advanced: context 8K · KV cache 16-bit
Context
KV cache

Smallest GPU for each quant and context (KV cache 16-bit)

Quantisation8K16K32K64K128K256K
Q3_K_M
Q4_K_M
Q5_K_M
Q6_K
Q8_0
FP16

≤16 GB card24–32 GB48–96 GBmore

QuantisationQualityWeightsKV cacheTotal
Q2_KHeavy lossSmallest, noticeably worse qualityHeavy loss55 GB est.0.8 GB≈59 GB
Q3_K_MNoticeable lossSmall, some quality lossNoticeable loss82 GB est.0.8 GB≈87 GB
Q4_K_MrecommendedSmall lossRecommended default for local useSmall loss102 GB est.0.8 GB≈108 GB
Q5_K_MNear-losslessNear-lossless for most tasksNear-lossless119 GB est.0.8 GB≈126 GB
Q6_KNear-losslessPractically losslessNear-lossless138 GB est.0.8 GB≈146 GB
Q8_0LosslessLossless in practice, twice the size of Q4Lossless175 GB file0.8 GB≈185 GB
FP16 / BF16LosslessFull precision, as releasedLossless330 GB file0.8 GB≈347 GB

Estimates for 8K tokens of context: weights + KV cache + runtime overhead. All experts stay in memory, but only the active ones are read per token. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it. Weights marked “file” use the real GGUF download size; “est.” rows are computed from the parameter count.

How to pick a quant

Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.

  • LosslessF16, Q8_0
  • Near-losslessQ6_K, Q5_K_M
  • Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
  • Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
  • Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K

Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B

How much context fits?

Model quant
KV cache

At 8K context, Q4_K_M needs ≈108 GB(102 GB weights + 0.8 GB KV cache + 5.6 GB overhead). Pick a GPU above to see how much context fits.

0501001501K2K4K8K16K32K64K128K256KGB

Weights + overheadKV cache

Can the Apple M2 Ultra (192 GB) run Qwen3.8-Flash-Next? →

Speed vs quality per quant

Smaller quants run faster and fit in less memory, but lose quality.

Q2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_0F16

Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.

Real-world speed of Qwen3.8-Flash-Next

Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →

No medians yet. Be one of the first 3 to report a setup.

All reports

Run Qwen3.8-Flash-Next locally

--quantQ4_K_Mdefault; pick your GPU to tailor itunsloth/Qwen3.8-Flash-Next-GGUF

From the Ollama librarydocs
ollama run qwen3.8-flash-next

The library build picks its own default quantisation.

UD-Q4_K_XL straight from Hugging Facedocs
ollama run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Install Ollama

OpenAI-compatible server with a web UIdocs
llama-server -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Downloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.

Install llama.cpp

Download with the lms CLIdocs
lms get unsloth/Qwen3.8-Flash-Next-GGUF@UD-Q4_K_XL

Then load it from the LM Studio app or with `lms load`.

Install LM Studio

Serve an OpenAI-compatible APIdocs
vllm serve Qwen/Qwen3.8-Flash-Next
Smaller FP8 builddocs
vllm serve Qwen/Qwen3.8-Flash-Next-FP8
Docker Compose (NVIDIA)docs
yaml
services:
  vllm:
    image: vllm/vllm-openai:latest
    command: Qwen/Qwen3.8-Flash-Next
    ports:
      - "8000:8000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 8000.

Install vLLM

Serve an OpenAI-compatible APIdocs
sglang serve Qwen/Qwen3.8-Flash-Next --host 0.0.0.0 --port 30000
Docker Compose (NVIDIA)docs
yaml
services:
  sglang:
    image: lmsysorg/sglang:latest
    command: sglang serve Qwen/Qwen3.8-Flash-Next --host 0.0.0.0 --port 30000
    ports:
      - "30000:30000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 30000.

Install SGLang

Ask about an imagedocs
python
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="Qwen/Qwen3.8-Flash-Next", dtype="auto", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
            {"type": "text", "text": "Describe this image."},
        ],
    }
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])

Install Transformers

Benchmarks

BenchmarkScoreRankSourceDate
GPQA Diamond91.7%#8 / 104 self-reportedAug 2026
Humanity's Last Exam35.9%#25 / 63 self-reportedAug 2026
SWE-bench Pro62.5%#5 / 51 self-reportedAug 2026
SWE-bench Multilingual81.0%#2 / 29 self-reportedAug 2026
LMArena WebDev1633#1 / 30 independentOct 2026

Self-reported scores come from the model card and are not independently verified. Rank is among all models with a score on that benchmark.

Quantised versions and downloads

RepositoryFormatQuantSizeDownloads
RadixArk/Qwen3.8-Flash-Next-NVFP4SafetensorsNVFP4—427.8K
nvidia/Qwen3.8-Flash-Next-NVFP4SafetensorsNVFP4—298.4K
Qwen/Qwen3.8-Flash-Next-FP8SafetensorsFP8—145.8K
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-IQ1_S73 GB1.2M
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-IQ1_M75 GB1.2M
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-Q2_K_XL79 GB1.2M
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-IQ3_XXS82 GB1.2M
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-Q3_K_XL90 GB1.2M
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-IQ4_XS94 GB1.2M
unsloth/Qwen3.8-Flash-Next-GGUFGGUFUD-Q4_K_XL111 GB1.2M

About Qwen3.8-Flash-Next

As the frontier of foundation models pushes toward ever-larger parameter counts and ever-longer context windows, the question is no longer just how much we can scale, but how efficiently we can do so. Sustainable progress toward artificial general intelligence (AGI) that benefits everyone demands architectural innovation. Today, we are sharing a concrete step in that direction: Qwen3.8-Flash-Next. This experimental preview of the architecture that will underpin Qwen4 is built around a fundamental rethinking of how the core components of modern large language models (LLMs) interact at scale. The first open-weight release under this architecture is Qwen3.8-Flash-Next, which introduces:

Full model card on Hugging Face →

Alternatives to Qwen3.8-Flash-Next

All alternatives →
zai-org logo

GLM-5.3-Flash

zai-org·Vision

≈213 GB

👋 Join our WeChat or Discord community. 📖 Check out the GLM-5.3-Flash blog and GLM-5 Technical report . 📍 Use GLM-5.3-Flash API services on Z.ai API Platform.

321BA18.3B1M ctx MIT
6.3MAug 2026ollama
thinkingmachines logo

Inkling-Small

thinkingmachines·Vision

≈159 GB

Inkling-Small is a general-purpose multimodal model that accepts text, image and audio inputs and generates text outputs. It is intended for use in English and other languages, and across multiple coding languages.

266BA1.7B1M ctx Apache 2.0
1.3MJul 2026gguf
MiniMaxAI logo

MiniMax-M3.1-Flash

MiniMaxAI·Vision

≈132 GB

MiniMax-M3.1-Flash is a native multimodal mixture-of-experts model for coding, agentic workflows, computer use, and cybersecurity. It supports text, image, and video inputs with a context window of up to 1M tokens.

220B1M ctx minimax-community
41.8KOct 2026ollama
Qwen logo

Qwen3.5-122B-A10B

Qwen·Vision

≈75 GB

Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.

125BA10B256K ctx Apache 2.0
942.5KFeb 2026ollama

NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 is the full-precision release of NVIDIA Nemotron 3.5 Super VL, a vision-language model for reasoning, coding, tool use, document understanding, and video understanding.

124BA12B1M ctx openmdw-1.1
68.3KOct 2026gguf
Qwen logo≈142 GB

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

236BA22B256K ctx Apache 2.0
323.5KSept 2025ollama

Qwen3.8-Flash-Next FAQ

How much VRAM does Qwen3.8-Flash-Next need?
About 108 GB at Q4_K_M with an 8K context. Q8_0 needs about 188 GB and full precision about 353 GB. Longer contexts need more memory for the KV cache.
Can Qwen3.8-Flash-Next run on the Apple M2 Ultra (192 GB)?
Yes. It fits at Q5_K_M (about 126 GB), at an estimated 10 tokens/s. It fits comfortably on the Apple M3 Ultra (512 GB).
Can I use Qwen3.8-Flash-Next commercially?
Yes, with conditions. It is released under qwen-community-1.0. Custom licence: read the model card before using it commercially.
What is the context length of Qwen3.8-Flash-Next?
262,144 tokens (256K). The full window needs far more memory than an 8K context.
How do I run Qwen3.8-Flash-Next with Ollama or llama.cpp?
With Ollama: ollama run qwen3.8-flash-next. With llama.cpp: llama-server -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL, which starts an OpenAI-compatible server.

Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.