Start a conversation. Inspect the response.
Chat directly with either prototype model on its independently configured worker. Recent context carries into your next message, responses stream as they are generated, and token usage is shown for each reply.
Conversation
16K contextSend a message to start a conversation. The real model’s replies will stream here.
The console is a prototype workspace.
MiniCPM uses a bounded CPU worker; Qwen3-8B uses a separate temporary L4 worker. The selected worker must be running for a live response. These integration checks are not model-quality benchmarks.
Limits follow the selected model.
MiniCPM has a 4K context and a 512-token output cap. Qwen3-8B has a 16K serving context and a 2,048-token output cap. Each chat includes up to four recent exchanges. Billing is not connected, and the API reference documents the remaining request limits.