Run Qwen 3.6 without a GPU: the hosted route and what it costs

Описание и применение

Пользователи могут запускать модель Qwen 3.6 35B A3B без GPU, используя хостинг LU Labs, оплачивая кредиты или подписку, с возможностью включения режима «think» и загрузки изображений.

Задача

Running the large Qwen 3.6 35B A3B model requires a GPU with ~24 GB VRAM, which many laptops, work machines, or Macs lack.

Результат

Users can run Qwen 3.6 35B A3B without a GPU by using LU Labs’ hosted service or API, paying per credit or a subscription, and optionally toggling reasoning or uploading images.

Рабочий процесс

• 1. Sign up at https://lu-labs.ai and open Studio.
• 2. Select Qwen 3.6 35B A3B from the hosted chat catalog.
• 3. (Optional) Toggle the "think" mode for multi‑step reasoning or disable it for quick answers.
• 4. (Optional) Upload images directly into the chat for vision tasks.
• 5. (API) Generate a key under Settings → Cloud API keys.
• 6. Call the endpoint https://lu-labs.ai/api/inference/v1/chat/completions with the model ID "Qwen/Qwen3.6-35B-A3B" and your messages.
• 7. Choose a payment plan: €19/month for 900,000 credits or a credit pack (€5 for 165,000 credits, €10 for 350,000 credits, €25 for 900,000 credits).
• 8. Use the credits to pay for inference; credits never expire on packs.
• 9. (Optional) For local use, install the free desktop app on a machine with a 24 GB GPU.
• 10. Monitor usage and costs via the LU Labs dashboard.

Доказательства

Qwen 3.6 35B A3B is in the base hosted chat catalog, the 15 models every hosted account reaches

Источник

Credits from one pool, and one credit is $0.00001

Источник

Hosted, €19 a month, 900,000 credits

Источник