OnPrem World地端世界
中文

Qwen3.8-Flash-Next

Qwen3.8-Flash-Next by Alibaba Qwen (180B parameters, MoE with 6B active per token). At Q4_K_M with an 8K context it needs about 107.4 GiB of memory. An experimental preview of Qwen's next architecture: a 125B MoE (6B active) plus a 51B n-gram embedding, 180B stored in all, so plan on a 192 GB-class machine and a recent runtime. Unlike the other Qwen models here, it uses the Qwen Community License rather than Apache-2.0.

Last updated: 2026-10-13

Model facts

DeveloperAlibaba Qwen
Parameters180B (6B active, MoE)[src]
Native context262,144 tokens[src]
LicenceQwen Community License 1.0 (qwen-community-1.0)[src]
Commercial useYes
KV cache12 of 48 layers (the rest are linear attention with a fixed-size state)[src]
Released2026-08

Good for: agents · coding · long context · Chinese · 128GB+ machines

Run it

ollama run qwen3.8-flash-next:125b-a6b-q4_K_M

Memory needed

Weights + KV cache + runtime overhead, in GiB. Lower quantization = less memory, slightly lower quality.

Quantization4K context8K context32K context
Q4_K_M107.3 GiB107.4 GiB108.0 GiB
Q5_K_M125.8 GiB125.9 GiB126.4 GiB
Q6_K145.6 GiB145.7 GiB146.2 GiB
Q8_0187.6 GiB187.7 GiB188.3 GiB
FP16352.6 GiB352.7 GiB353.3 GiB

Which hardware runs it

Best quantization that fits each device with a 8K context. Estimates, not benchmarks. See the methodology.

HardwareMemoryBest quanttok/s (est.)
M5 Ultra512 GBFP1665
M3 Ultra512 GBFP1644
Ryzen AI Max+ PRO 495 (Gorgon Halo) systems192 GBQ5_K_M38
GeForce RTX 3060 (12 GB)12 GBDoesn't fit–
GeForce RTX 309024 GBDoesn't fit–
GeForce RTX 4060 Ti 16GB16 GBDoesn't fit–
GeForce RTX 4080 SUPER16 GBDoesn't fit–
GeForce RTX 409024 GBDoesn't fit–
GeForce RTX 5060 Ti 16GB16 GBDoesn't fit–
GeForce RTX 507012 GBDoesn't fit–
GeForce RTX 5070 Ti16 GBDoesn't fit–
GeForce RTX 508016 GBDoesn't fit–
GeForce RTX 509032 GBDoesn't fit–
Radeon RX 7900 XT20 GBDoesn't fit–
Radeon RX 7900 XTX24 GBDoesn't fit–
Radeon RX 9060 XT 16GB16 GBDoesn't fit–
Radeon RX 9070 XT16 GBDoesn't fit–
Arc B58012 GBDoesn't fit–
RTX A600048 GBDoesn't fit–
RTX 6000 Ada Generation48 GBDoesn't fit–
RTX PRO 5000 72GB Blackwell72 GBDoesn't fit–
RTX PRO 6000 Blackwell Workstation Edition96 GBDoesn't fit–
Radeon AI PRO R970032 GBDoesn't fit–
Arc Pro B6024 GBDoesn't fit–
Arc Pro B7032 GBDoesn't fit–
M4 Pro64 GBDoesn't fit–
M4 Max (14-core CPU / 32-core GPU)36 GBDoesn't fit–
M4 Max (16-core CPU / 40-core GPU)128 GBDoesn't fit–
M532 GBDoesn't fit–
M5 Pro64 GBDoesn't fit–
M5 Max (18-core CPU / 32-core GPU)36 GBDoesn't fit–
M5 Max (18-core CPU / 40-core GPU)128 GBDoesn't fit–
M6 (16GB configuration, 153 GB/s)16 GBDoesn't fit–
M6 (24GB/32GB configurations, 170 GB/s)32 GBDoesn't fit–
DGX Spark128 GBDoesn't fit–
RTX Spark (N1X) laptops & compact desktops128 GBDoesn't fit–
Ryzen AI Max+ 395 (Strix Halo) systems128 GBDoesn't fit–

Frequently asked questions

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

With an 8K context, about 107.4 GiB at Q4_K_M and 187.7 GiB at Q8_0 (weights + KV cache + runtime overhead). Longer contexts need more memory for the KV cache.

Can Qwen3.8-Flash-Next run on an GeForce RTX 4090?

Qwen3.8-Flash-Next does not fit on the GeForce RTX 4090 (24 GB), even at Q4_K_M with an 8K context.

Can I run Qwen3.8-Flash-Next on a Mac?

Yes. Qwen3.8-Flash-Next fits on the M3 Ultra (256 GB) at Q8_0 with an 8K context, using about 187.7 GiB, at an estimated 84 tokens/s.

Can Qwen3.8-Flash-Next be used commercially?

Qwen3.8-Flash-Next is released under Qwen Community License 1.0 (qwen-community-1.0). The licence allows commercial use (read its conditions).

Sources