Models · Qwen

Deploy Qwen models on dedicated GPUs

56 Qwen models ready to deploy on GPU capacity reserved for you, from 596M to 2.45T parameters, covering chat & reasoning, coding, vision & multimodal, embeddings, rerankers, speech recognition, text-to-speech and safety & moderation. Private OpenAI-compatible endpoint, European GPUs, billed per GPU-hour.

Request a deployment ↓Qwen on Hugging Face ↗All makersHow dedicated works
ModelParams
Qwen3.8-27Btrendingvision
Qwen		28B
Qwen3.8-Flash-NexttrendingvisionMoE
Qwen		180B
Qwen3.6-35B-A3BtrendingvisionMoE
Qwen		36B
Qwen3.5-9Btrendingvision
Qwen		9.7B
Qwen3-TTS-12Hz-1.7B-CustomVoicetrending
Qwen		1.9B
Qwen3-8B
Qwen		8.2B
Qwen3-0.6B
Qwen		752M
Qwen3.5-4Btrendingvision
Qwen		4.7B
Qwen3-Embedding-0.6B
Qwen		596M
Qwen3.6-27Bvision
Qwen		28B
Qwen3-ASR-1.7Btrending
Qwen		2.3B
Qwen3-VL-8B-Instructvision
Qwen		8.8B
Qwen3.5-35B-A3BvisionMoE
Qwen		36B
Qwen3-4B-Instruct-2507
Qwen		4.0B
Qwen3-Coder-NextMoE
Qwen		80B
Qwen3-Coder-30B-A3B-InstructMoE
Qwen		31B
Qwen2.5-Omni-7Bvision
Qwen		11B
Qwen3.5-27Bvision
Qwen		28B
Qwen3.8-2.4T-A95BMoE
Qwen		2.45T
Qwen3-14B
Qwen		15B
Qwen3-Embedding-8B
Qwen		7.6B
Qwen3.5-0.8Bvision
Qwen		873M
Qwen3-1.7B
Qwen		2.0B
Qwen3-30B-A3BMoE
Qwen		31B
Qwen3-235B-A22BMoE
Qwen		235B
Qwen3-Omni-30B-A3B-Instructvision
Qwen		35B
Qwen3.5-122B-A10BvisionMoE
Qwen		125B
Qwen3.5-2Bvision
Qwen		2.3B
Qwen3-VL-Embedding-8B
Qwen		8.1B
Qwen3-Reranker-0.6B
Qwen		596M
Qwen3-Next-80B-A3B-InstructMoE
Qwen		81B
Qwen3-VL-30B-A3B-InstructvisionMoE
Qwen		31B
Qwen3-30B-A3B-Instruct-2507MoE
Qwen		31B
Qwen3.5-397B-A17BvisionMoE
Qwen		403B
Qwen3-ASR-0.6B
Qwen		938M
QwQ-32B
Qwen		33B
Qwen3-VL-Embedding-2B
Qwen		2.1B
Qwen3-32B
Qwen		33B
Qwen3-Coder-480B-A35B-InstructMoE
Qwen		480B
Qwen3-VL-4B-Instructvision
Qwen		4.4B
Qwen2.5-VL-72B-Instructvision
Qwen		73B
Qwen3-4B-Thinking-2507
Qwen		4.0B
Qwen3-VL-2B-Instructvision
Qwen		2.1B
Qwen3-235B-A22B-Instruct-2507MoE
Qwen		235B
Qwen3-Reranker-4B
Qwen		4.0B
Qwen2.5-VL-32B-Instructvision
Qwen		33B
Qwen3-VL-235B-A22B-InstructvisionMoE
Qwen		236B
Qwen3-30B-A3B-Thinking-2507MoE
Qwen		31B
Qwen3-VL-Reranker-2B
Qwen		2.1B
Qwen3-Embedding-4B
Qwen		4.0B
Qwen3-Next-80B-A3B-ThinkingMoE
Qwen		81B
Qwen3Guard-Gen-8B
Qwen		8.2B
Qwen3-Reranker-8B
Qwen		8.2B
Qwen3-VL-235B-A22B-ThinkingvisionMoE
Qwen		236B
Qwen3-235B-A22B-Thinking-2507MoE
Qwen		235B
Qwen3-VL-Reranker-8B
Qwen		8.8B

Specs from Hugging Face, synced 2026-10-27. Machine-readable version: https://sference.com/models/Qwen.md

FAQ

Which Qwen models can I deploy on sference?

56 today, most requested first: Qwen3.8-27B, Qwen3.8-Flash-Next, Qwen3.6-35B-A3B, Qwen3.5-9B, Qwen3-TTS-12Hz-1.7B-CustomVoice, Qwen3-8B, Qwen3-0.6B and Qwen3.5-4B, and 48 more. Any other Qwen checkpoint or a fine-tune of one can be requested with the form on this page.

How many GPUs do Qwen models need?

It depends on the model: the lineup runs from 596M to 2.45T parameters. Each model page lists the checkpoint size and the minimum footprint on 80 GB and 141 GB GPUs. You don't pick hardware on sference: you choose a performance tier and we size the deployment.

Can I use Qwen models commercially?

53 of the 56 listed models are under licenses that allow commercial use (Apache 2.0 and Custom license across the lineup). Each model page summarizes its license and links the full text.

How is a dedicated deployment priced?

Per GPU-hour on a monthly commitment, sized to your throughput and latency targets.

Request a Qwen model

Tell us which Qwen model and roughly how much traffic you expect. Fine-tunes of these models work the same way.

  1. 01
    We size it

    A performance tier, the capacity behind it, and a quote for your volume.

  2. 02
    We benchmark it

    Bring one workload with a sample of real traffic; we come back with latency and throughput numbers.

  3. 03
    It goes live

    We provision the deployment. You call it with the same API and model id as before.

No commitment. Prefer email? [email protected]