Model catalogQwen
Qwen3.8-Flash-Next
Qwen4-architecture preview — ultra-sparse, fast, and multimodal. Served with vLLM behind your own private, OpenAI-compatible /v1 endpoint — billed by the second, terminate anytime.
125B MoE · 6B act · FP8 · 256K context · 4× 96 GB · per-second billing
01Specifications
The specs the launcher enforces.
Architecture, served precision, max context, the minimum GPU shape, and the exact pinned weights — the facts that decide how Qwen3.8-Flash-Next runs and what it costs.
| Spec | Value |
|---|---|
| Architecture | 125B MoE · 6B act |
| Served precision | FP8 |
| Max context | 256K |
| Minimum GPU shape | 4× 96 GB |
| Weights served | Qwen/Qwen3.8-Flash-Next-FP8@970c569 |
Runs on 4× 96 GB · billed per second — final cost depends on the GPU shape.
Full pricing →02How it runs
A private endpoint, not a shared API.
Launching Qwen3.8-Flash-Next runs the model on your own instance and gives you a private, OpenAI-compatible /v1 endpoint and API key. Point any OpenAI-compatible SDK at your base URL, change the model name, and existing code works. The endpoint is yours alone — terminate the instance and it goes away with it.
Billing is by the second on the 4× 96 GB shape the launcher enforces, from boot to terminate. No commitment and no tier gates.
03Get started
Launch Qwen3.8-Flash-Next now.
One private OpenAI-compatible endpoint, one key, billed by the second. The console opens with Qwen3.8-Flash-Next preselected.