Bittensor's Subnet 92 claims a 'frontier JEV model' 200x faster than LLMs
markjeffrey · x · 2026-11-23
markjeffrey amplifies the claim that Bittensor ($TAO) Subnet 92 (Reax) has produced the world's top 'frontier JEV model' — a new class 200x faster than LLMs at snap decisions like re-ordering 1000 support tickets by severity. He estimates 20-30% of current LLM prompts could be offloaded to such 'math coprocessor' decision models. Claims are unverified.
More from Infra
- Only 2% of announced 90GW self-powered data center capacity in US actually running — rohanpaul_ai · 2026-11-23
- Vercel CEO Rauch: 'free' ZDR stealth models look like industrial-scale data exfiltration — AccBalanced · 2026-11-23
- Lenovo's enterprise AI server sales surge 200%, outpacing cloud giants' 170% — shashib · 2026-11-23
- Australia's Spy Agency Now Treats Data Center Ownership as Foreign Interference Risk — rohanpaul_ai · 2026-11-23
- RTX Spark 18C vs M5 Pro/Max Benchmarks: 32% Faster Prompt Processing, But Apple Wins on Decode — Temporary-Size7310 · 2026-11-23
- RustingStrata 1.3.0: a Rust GGUF engine that runs models bigger than your VRAM — Serious_Scarcity7818 · 2026-11-23