Hugging FaceAn early look at the design behind Qwen4, and it reads pictures and thinks before it answers.
- Rating
- 4.0
- Downloads
- 9.4M
- Parameters
- 177B, 6B active
- Context
- 256K
- Download
- 73 to 355 GB
- Commercial use
- With conditions
Too big for this Mac
Its smallest size needs 91 GB of memory, more than M4 · 16 GB can give it.
About
It is the first open release built on this new Qwen design.
How it compares
Of the 6 chat models Outmake recommends for a 128 GB Mac, it was the 2nd most downloaded on Hugging Face last month.
Outmake estimates it is the 3rd fastest of those six, and Qwen3.6 35B writes about 1.4 times as fast.
Of those six, Outmake estimates it writes about 2.4 times as fast as MiniMax M2.7, the one that can't read pictures.
Sources: Qwen/Qwen3.8-Flash-Next model card, Outmake's speed estimates, LM Arena and Hugging Face. Every line comes from one of these.
Good at
Licence
Qwen Community License 1.0. Commercial use: with conditions. Commercial use is allowed, but a product or service with more than 100 million monthly active users or $20 million in monthly revenue must display the model name on its interface, and running it as a hosted Model-as-a-Service or AI Work Assistant business needs a separate license from Qwen. The maker's page has the full terms.
Qwen3.8 Flash Next runs well from a Mac with an M1 Ultra and 128 GB of memory, at UD-Q3_K_XL, writing about 28 words (37 tokens) a second. 10 of 59 Mac configurations run it well.
Compare with
The nearest models in size, judged for the same Mac.
- MiniMax M2.7Help finding bugs in codeToo big for this Mac · 102 GBGet
- Qwen3.5 122BFor long documents and questions in many languages, and it can read photos and screenshots too.Too big for this Mac · 58 GBGet
- GPT-OSS 120BOpenAI's open-weight reasoning model, designed for reasoning problems and agent tasks.Too big for this Mac · 63 GBGet
- DeepSeek V4 FlashLong documents and tool useToo big for this Mac · 128 GBGet
Sizes
| Size | Download | Memory at 8K | Memory at 32K | Open |
|---|---|---|---|---|
| UD-Q3_K_XLSmaller | 91 GB | 62 GB | 63 GB | Open |
| UD-Q4_K_XLBalanced | 112 GB | 84 GB | 84 GB | Open |
| UD-Q6_K_XLBest | 170 GB | 116 GB | 116 GB | Open |
All 11 sizes
| Size | Download | Memory at 8K | Memory at 32K | Open |
|---|---|---|---|---|
| UD-IQ1_S | 73 GB | 73 GBestimated | 73 GB | Open |
| UD-IQ1_M | 75 GB | 75 GBestimated | 75 GB | Open |
| UD-Q2_K_XL | 80 GB | 80 GBestimated | 80 GB | Open |
| UD-IQ3_XXS | 83 GB | 83 GBestimated | 83 GB | Open |
| UD-IQ4_XS | 95 GB | 95 GBestimated | 95 GB | Open |
| UD-Q5_K_XL | 159 GB | 159 GBestimated | 159 GB | Open |
| Q8_0 | 189 GB | 189 GBestimated | 189 GB | Open |
| BF16 | 355 GB | 355 GBestimated | 355 GB | Open |
Memory is what llama.cpp's own calculator reported for each size when the catalog was built, with the app's settings. Smaller, Balanced and Best are the three sizes the store names.
Run It Anywhere
Qwen3.8 Flash Next is an open model, so any app that runs GGUF files can run it. These take the UD-Q4_K_XL size from unsloth straight from Hugging Face.
- LM StudioMac, Windows, Linux
Open - JanMac, Windows, Linux
Open - MstyMac, Windows, Linux
Open - OllamaMac, Windows, Linux
ollama run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL - llama.cppMac, Windows, Linux
llama-server -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL - Docker Model RunnerMac, Windows, Linux
docker model run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
On a PC Graphics Card
| Video memory | Largest size that fits | Download | Memory at 8K |
|---|---|---|---|
| 6 GB | None worth running | needs 62 GB | |
| 8 GB | None worth running | needs 62 GB | |
| 10 GB | None worth running | needs 62 GB | |
| 11 GB | None worth running | needs 62 GB | |
| 12 GB | None worth running | needs 62 GB | |
| 16 GB | None worth running | needs 62 GB | |
| 20 GB | None worth running | needs 62 GB | |
| 24 GB | None worth running | needs 62 GB | |
| 32 GB | None worth running | needs 62 GB | |
| 48 GB | None worth running | needs 62 GB |
The largest size, of those that keep three bits a weight or more, whose memory at an 8K chat is under 92% of the card's; the rest is the desktop's. Estimated from the files, the same way as for a Mac; speed on a PC is not estimated yet. A size that doesn't fit the card still runs, slower, with part of it in ordinary memory.
Tests
GPQA Diamond: questions PhD scientists write for each other. Humanity's Last Exam: questions written to be beyond today's models. Scores are the ones filed on the model's Hugging Face page or the board's own table, read 2026-11-05; some come from the maker.
On every Mac
| Mac | Size | Fit | Memory | Writes | Reads |
|---|---|---|---|---|---|
| M6 · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M6 · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M6 · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Pro · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Pro · 48 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Pro · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Max · 36 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Max · 48 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Max · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M5 Max · 128 GB | UD-Q3_K_XL | Runs well | 91 GB | 46 | 1354 |
| M5 Ultra · 96 GB | UD-IQ1_M | Tight fit | 75 GB | 50 | 2167 |
| M5 Ultra · 256 GB | UD-Q6_K_XL | Runs great | 170 GB | 54 | 2709 |
| M5 Ultra · 512 GB | UD-Q6_K_XL | Runs great | 170 GB | 54 | 2709 |
| M4 · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Pro · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Pro · 48 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Pro · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Max · 36 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Max · 48 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Max · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M4 Max · 96 GB | UD-IQ1_M | Tight fit | 75 GB | 16 | 364 |
| M4 Max · 128 GB | UD-Q3_K_XL | Runs well | 91 GB | 28 | 455 |
| M3 · 8 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 Pro · 18 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 Pro · 36 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 Max · 36 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 Max · 48 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 Max · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M3 Max · 96 GB | UD-IQ1_M | Tight fit | 75 GB | 13 | 283 |
| M3 Max · 128 GB | UD-Q3_K_XL | Runs well | 91 GB | 25 | 377 |
| M3 Ultra · 96 GB | UD-IQ1_M | Tight fit | 75 GB | 27 | 565 |
| M3 Ultra · 256 GB | UD-Q6_K_XL | Runs great | 170 GB | 29 | 565 |
| M3 Ultra · 512 GB | UD-Q6_K_XL | Runs great | 170 GB | 29 | 753 |
| M2 · 8 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 · 24 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 Pro · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 Pro · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 Max · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 Max · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 Max · 96 GB | UD-IQ1_M | Tight fit | 75 GB | 16 | 351 |
| M2 Ultra · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M2 Ultra · 128 GB | UD-Q3_K_XL | Runs well | 91 GB | 37 | 555 |
| M2 Ultra · 192 GB | UD-Q4_K_XL | Runs great | 112 GB | 34 | 555 |
| M1 · 8 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 Pro · 16 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 Pro · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 Max · 32 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 Max · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 Ultra · 64 GB | UD-Q3_K_XL | Too big | 91 GB | ||
| M1 Ultra · 128 GB | UD-Q3_K_XL | Runs well | 91 GB | 28 | 411 |
Writes and reads are words a second (a token is about three quarters of a word), 4K tokens into a chat, estimated unless marked measured. The size is the one the store offers that Mac. Constants fitted October 11, 2026; how, and how wrong.
Badge for the model card
Shows on Qwen3.8 Flash Next's Hugging Face card which Macs run it well, linked to this page. It updates every morning.
Sizes and files are read from unsloth/Qwen3.8-Flash-Next-GGUF on Hugging Face and checked every morning. LM Arena has not rated this model, so the rating is Outmake's estimate from its size and release date; how the estimate is made. Catalog of November 4, 2026.