# Edge0 releases 145B and 120B models with local inference tools

_Published Thursday, November 12, 2026 at 8:06 PM EST · AI · Latest · Tier 2 — Notable_

![Edge0 releases 145B and 120B models with local inference tools — Primary](https://cdn.hashnode.com/uploads/covers/6aea03219fa4aa99923b5acf/de5ebdea-3436-4b9a-b3f2-e9910f3a0abc.png)

Edge0 says it has open-sourced its Edge0-145B and Edge0-120B models, inference framework and deployment tools for local coding, reasoning and agent workflows. Day-one support covers macOS, Windows and Linux, including Apple M-series chips and specified NVIDIA platforms.

The company reports 29GB peak memory and decoding above 80 tokens per second for Edge0-145B on compatible laptop hardware. Its design shares matrices across groups of experts and overlaps computation with access to GPU memory, system memory and SSD storage. Performance and several benchmark comparisons come from internal testing; comparison models' memory footprints are estimates rather than measured results.

## Sources

- [edge0.ai](https://edge0.ai/blog/introducing-edge0-145b-frontier-ai-capabilities-on-your-own-computer)

---
Canonical: https://techandbusiness.org/newswire/2T53E7K1V8XNoGTr_v-9th
Published: 2026-11-13T01:06:41.549Z
Story chronology: 2026-11-12T18:00:00.000Z
Retrieved: 2026-11-13T03:04:03.453Z
Publisher: Tech & Business (techandbusiness.org)