The request names qwen3.5 and a tiny test model for the Qwen3.5 conditional-generation architecture. Replicate hosts open-source models through a cloud API, and Ollama runs open AI models locally or in the cloud, so both are the most direct ways to run Qwen-family models. Hugging Face hosts models and offers SDK and CLI access to them, and Modal runs model inference at scale.

1. 1

   Replicatereplicate.comverified and signed

   Runs open-source machine learning models such as Qwen models through a cloud API, and its action record lists qwen/qwen3-235b-a22b-instruct-2507, an updated Qwen3 instruction-following model 11.

   WebsiteDocsPricingGet started

   +14 community tools built for it

   for agents: lucataco/qwq-32b · openai/o1-mini · qwen/qwen3-235b-a22b-instruct-2507

2. 2

   Ollamaollama.comwebsite read, tools not yet checked

   Runs open AI models locally or in the cloud on macOS, Windows and Linux, which suits running a small Qwen3.5 test model 6.

   WebsiteDocsPricingGet started

   +141 community tools built for it

3. 3

   Hugging Facehuggingface.coverified and signed

   Hosts models and datasets and provides the huggingface-hub SDK and CLI tooling to fetch and run them 2.

   WebsiteDocsPricingGet started

   official:MCP server · @huggingface/mcp-clientSDK · huggingface-hubCLI · text-embeddings-inferencerepository · pwc-cliskill · hf-cli+23 community tools built for it

   for agents: tiny-agent · agent · llm

4. 4

   Modal: The platform for production AImodal.comverified and signed

   Provides cloud infrastructure to serve AI models at scale with OpenAI-compatible inference endpoints 9.

   WebsiteDocsPricingGet started

   +20 community tools built for it

   for agents: serverless · modal · cloud