OpenAI-compatible embeddings: Qwen3-VL-Embedding-2B (2048 dims), text and images, on Watt's own fleet. Pay per call in USDC; the price is the request's tokens, estimated before it is served.
OpenAI-compatible chat completions: Qwen3.5 4B, Qwen3 1.7B and 0.6B on Watt's own fleet. Pay per call in USDC; the price is the messages' tokens plus max_tokens, estimated before it is served.