W&B Inference by Weights & Biases

W&B Inference, powered by CoreWeave, offers API and playground access to leading open-source LLMs like DeepSeek, Qwen3, Kimi K2, Llama 4, Phi, and OpenAI GPT OSS within the Weights & Biases platform. It allows users to develop AI applications without hosting models, juggling API keys, or managing clusters. Users can call models through a unified API, test and switch between models, build AI agents, and monitor usage with W&B Weave. Inference runs on CoreWeave’s GPU infrastructure, integrates with the W&B ecosystem, and provides a predictable cost plan without needing separate provider accounts.