Edge0 open sources 145B and 120B edge-side models and inference frameworks.

Edge0 has open-sourced two edge models, Edge0-145B and Edge0-120B, along with a unified inference framework and deployment tools. Edge0 claims the 145B model can run locally on compatible laptops, with peak memory of 29GB, decoding speed of 80+ tokens/s, and prefill speed of 1100+ tokens/s, supporting 128K–256K context. Its core capabilities can be used offline without accounts or usage quotas. Edge0 claims the 145B model supports macOS, Windows, and Linux on the first day, covering Apple M series chips and NVIDIA platforms such as RTX Spark, RTX 4090, and RTX 5090. Low-power devices currently cannot run the 145B model locally. The above runtime data and support range are Edge0's statements.

AI-written from the reports · Updated 18 hours ago