CoreWeave Forge
CoreWeave's unified development layer connecting model/agent inference, observability, data curation, training, and evaluation in one loop.
Overview
CoreWeave announced Forge on September 30, 2026 at its Fully Connected conference, a development layer that runs the full AI loop -- run, observe, curate, improve, evaluate, repeat -- in one connected environment while staying open to models, frameworks, and clouds teams already use outside CoreWeave Cloud. Forge unifies Weights & Biases Models, post-training expertise from OpenPipe, and the open-source marimo notebook project with CoreWeave's own inference and training services into a single stack for production observability, data curation, training, and evaluation. MasterClass and Canva are early adopters building on the platform.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- Connects model and agent inference, production observability, data curation, training, and evaluation in one environment
- Unifies Weights & Biases Models and OpenPipe post-training tooling with CoreWeave's own infrastructure
- Built on the open-source marimo notebook project
- Works across models, frameworks, and cloud environments beyond CoreWeave Cloud
Use cases
- Closing the loop between production agent/model runs and the next training or evaluation cycle
- Curating production data for post-training without leaving the observability stack
- Running evaluation and observability across models hosted outside CoreWeave's own cloud
Pricing
unknown -- not disclosed at launch; positioned for enterprise AI development teams
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 30 Sept 2026
- Record updated
- 1 Oct 2026
- Record last checked
- 1 Oct 2026