Anthropic Open-Sources FlashProve, Zero-Knowledge Proofs of AI Inference, Tested on Llama 4 Maverick
yenkel · x · 2026-11-03
Anthropic released FlashProve, an open-source library for provable inference: it lets an AI provider prove which model answered a request without revealing model weights.
- Benchmarked on Llama 4 Maverick, a 400B-parameter MoE model
- Code and blog post are public; repo is labeled a research prototype (Apache 2.0), not maintained
- Could enable verifiability and security-audit scenarios; yenkel notes interesting possibilities if prompts and proofs can be passed around
More from Research
- Mechinterp's unsung root: nearly all weight-editing work traces back to one VGEL post — max_paperclips · 2026-11-03
- NVIDIA's AgentBaton cuts agent API costs 65.9% via well-timed model handoffs — amaarora · 2026-11-03
- SJTU-led paper proposes 'Avatar Intelligence' and an Agentic Oversight Index for personal agents — SJTU · 2026-11-03
- Vapor-cell atomic E-field sensor hits 0.2 mV/m/√Hz in 11 mm³, no electrodes — jwt0625 · 2026-11-03
- BiGym 2.0: Claude-written policies hit 51.9% on humanoid household tasks, nearing VLA — teortaxesTex · 2026-11-03
- Hopfield model thermodynamic limit proven at all loads without replica-symmetry assumptions — _onionesque · 2026-11-03