Embedl opens free browser-based Jetson testing: INT4 quantization runs Cosmos3-Edge on 8GB Orin Nano
No-Dragonfly6246 · reddit · 2026-10-23
Embedl's device farm (embedl.com) lets you test Hugging Face models on real Jetson boards from the browser for free — pick a device, type a prompt or upload an image, and get runtime stats; 1000+ models supported. Highlight: Cosmos3-Edge didn't fit on Orin Nano Super 8GB in BF16, but INT4 quantization brings peak VLM memory to 3.3GB, hitting 152.5 tok/s on Orin Nano Super and 207.8 tok/s on AGX Thor 128GB (vs 44.1/68.2 tok/s with BF16 vLLM), while accuracy holds across five VLM benchmarks (65.56 → 65.20). VLM part is live; diffusion and policy components coming soon.
Related event: Embedl Opens Free Browser-Based Jetson Testing Platform(2 posts)→
More from Infra
- Which US State Will Rebrand as 'The Data Center State' and Build a Sovereign Wealth Fund? — matt_slotnick · 2026-10-23
- Qdrant benchmarks Cohere Embed 5 Pro: one dense vector per PDF page vs ColPali — qdrant_engine · 2026-10-23
- Owning GPUs Doesn't Make You Sovereign, Argues Air Street Capital's Nathan Benaich — nathanbenaich · 2026-10-23
- DeepInfer's math shows trailing SOTA maximizes profit under compute constraints — examachine · 2026-10-23
- NVIDIA deliberately halved FP32 accumulation on consumer RTX GPUs — unlocks possible for 20/30 series, not 50 — panchovix · 2026-10-23
- Benchmark: Direct gcloud CLI beats Skills and MCP for token-cheap agent deploys — rseroter · 2026-10-23