The cockpit for private AI work.
Arena is the visible cockpit. Cortex gives it local memory and provenance. fieldkit carries the reusable patterns around it. Together they make private AI work inspectable instead of mysterious.
The Arena cockpit on a real run: local models side by side, with live speed and memory telemetry. Nothing you type leaves your machine.
Compare
Run models side by side, track speed and memory, and keep a local leaderboard.
Remember
Index notes and documents with source trails so recall can be tested, not guessed.
Evaluate
Score behavior, benchmarks, and regressions before a workflow is trusted.
Ship
Turn working local patterns into repeatable artifacts and playbooks.
Built on the NVIDIA stack
Orionfold is a member of the NVIDIA Inception program. The Cockpit stack is proven on NVIDIA DGX Spark and supports NVIDIA GPU acceleration for inference, fine-tuning, and model evaluation workflows.
See the DGX Spark field lab →© 2025 NVIDIA, the NVIDIA logo, NVIDIA DGX Spark, NVIDIA Inception, NVIDIA NeMo, NVIDIA NIM, and TensorRT-LLM are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
