Edge0 open-sources streaming MoE framework with 35B and 8B on-device models
EAccelerate_42 · x · 2026-09-10
Edge0 is now open source, shipping 35B and 8B models plus local runtime and inference code built for phones, laptops, PCs and robots — fast, private, fully on-device. The streaming MoE framework combines SSD expert offload, Recover-LoRA and prerouter routing prediction, with a decoupled backend design: the current MLX backend targets Apple Silicon, and CUDA and other platforms plug into the same core abstractions. edge0-35B-A3B runs 4-bit with 40 layers and 256 experts (prerouter K=4); edge0-8B-A1B runs 4-bit with 24 layers and 128 experts (K=8). Checkpoints, LoRA adapters and prerouter heads are released as end-to-end units. The repo is new, with 88 stars so far.
Related event: Edge0 Open-Sources On-Device MoE Framework with 35B Model(2 posts)→
More from Infra
- Anthropic hits $16.3B H1 revenue as compute efficiency jumps 5.2x in 18 months — daniel_mac8 · 2026-10-27
- Tenstorrent runs Proteina protein model at high throughput, claims no HBM shortage — gajesh · 2026-10-27
- YC-backed Hopper launches voice-agent inference cloud: 80 ms TTFT at $0.30/1M tokens — ycombinator · 2026-10-27
- A100 rental prices still rising in 2026 per a16z data, Jevons paradox hits compute — VraserX · 2026-10-27
- Anthropic S-1 reveals $800B+ in off-books future compute commitments — GaryMarcus · 2026-10-27
- ~42% of Anthropic's ~11GW contracted compute is NVIDIA GPUs — ns123abc · 2026-10-27