{"slug": "anthropic-says-flashprove-proves-model-identity-at-70-to-95-times-inference-gpu", "title": "Anthropic says FlashProve proves model identity at 70 to 95 times inference GPU time", "summary": "Anthropic published FlashProve on November 2nd, an open-source research prototype that attaches a cryptographic proof to an AI response so a recipient can verify which committed model produced it, tested at 400B-parameter scale. The proof adds 70 to 95 times inference GPU time and is limited to 1,024 tokens, and Anthropic said it has no plans to deploy FlashProve in Claude. The prototype shows cryptographic receipts for model identity are technically feasible at 400B scale, while the compute overhead and token limit explain why no provider has deployed them yet.", "body_md": "# Anthropic says FlashProve proves model identity at 70 to 95 times inference GPU time\n\n**Anthropic's November 2nd research prototype tested proofs on a 400B model, with no plans to deploy it in Claude.**\n\n        By [RuntimeWire Staff](https://runtimewire.com/author/runtimewire-staff)\n        · Published \n        · Updated \n\nPrimary source: [TechCrunch](https://techcrunch.com/2026/11/03/anthropic-is-working-on-a-system-to-prove-ai-models-identity/)\n\n## Why it matters\n\nHosted AI buyers currently have to trust that the model they paid for is the model that answered. FlashProve shows cryptographic receipts for model identity are technically possible at 400B scale, while its 70 to 95 times compute overhead and 1,024-token limit explain why no provider has deployed them yet.\n\nAnthropic published [FlashProve](https://www.anthropic.com/research/provable-inference?ref=runtimewire) on November 2nd, an open-source prototype that attaches a cryptographic proof to an AI response so a recipient can check which committed model produced it. The work fits the thesis that brought CEO Dario Amodei and President Daniela Amodei out of OpenAI. Anthropic was founded in 2021 by former OpenAI members, including the Amodei siblings, who serve as CEO and president respectively. Dario Amodei had been OpenAI's vice president of research.…\n\nWhen you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.", "url": "https://wpnews.pro/news/anthropic-says-flashprove-proves-model-identity-at-70-to-95-times-inference-gpu", "canonical_source": "https://runtimewire.com/article/anthropic-flashprove-provable-inference-model-identity", "published_at": "2026-11-03 23:12:09+00:00", "updated_at": "2026-11-04 02:19:21.853308+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-safety", "ai-research", "large-language-models"], "entities": ["Anthropic", "FlashProve", "Claude", "Dario Amodei", "Daniela Amodei", "OpenAI", "TechCrunch"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/anthropic-says-flashprove-proves-model-identity-at-70-to-95-times-inference-gpu", "markdown": "https://wpnews.pro/news/anthropic-says-flashprove-proves-model-identity-at-70-to-95-times-inference-gpu.md", "text": "https://wpnews.pro/news/anthropic-says-flashprove-proves-model-identity-at-70-to-95-times-inference-gpu.txt", "jsonld": "https://wpnews.pro/news/anthropic-says-flashprove-proves-model-identity-at-70-to-95-times-inference-gpu.jsonld"}}