How fast anthropics/FlashProve is gaining stars, and where it lands on Trendshift and GitHub Trending
Activity and trending
Stars, forks and merged pull requests each day or month, lined up with the repository's rank on Trendshift Trending
Nov 2 – Nov 4, 2026
Stars +558in 3 days
0200400
343
Forks +0in 3 days
01
Merged PRs 0in 3 days
01
Trendshift Trendingwhere the repository ranked each day it made the top 25
Not in the Trendshift top 25 in this period
Nov 2Nov 3Nov 4
The left and right arrow keys step from day to day; Home and End go to the first and the last
Stars, forks, merged pull requests and Trendshift Trending rank, by day
Day
Stars gained
Forks gained
Merged PRs
Mon, Nov 2, 2026
+0
+0
0
Tue, Nov 3, 2026
+343
+0
0
Wed, Nov 4, 2026
+215
+0
0
Total
+558
+0
0
1#1 on the list
7Rank that day
Not trending
Social mentions
Recent discussions about this repository across the web
5 posts8 posts
The part the headline skips: Anthropic says Claude built it. The security claims are proved in Lean, so people can check its work. On a 400B model, each proof is 19 MB and checks one token. A CPU…
Anthropic published FlashProve, an open-source (Apache 2.0) prototype for cryptographic proofs of LLM inference. Per Anthropic, on Llama 4 Maverick (400B MoE, 4-bit) one H100 proves ~110 tokens/s; a…
You used to take the vendor's word that the named model answered. Nov 2, 2026: Anthropic open-sources FlashProve — a zkSNARK so a reply can prove it came from one committed weight set, without…
The old way to prove which AI model wrote a reply made a 93.6 GB proof that took almost a day to check. Anthropic's new FlashProve makes a 19 MB proof that a CPU checks in 17 seconds. Same 400B…
Anthropic put out FlashProve, an open-source way for a model provider to prove which model actually answered your request, without showing you the weights the problem is real tbh. right now you can't…
anthropic made this tweet 3 hours ago and linked their github: which if you look at the employees' names it all ends in -ant the official cult of anthropic ahead of their IPO
New @AnthropicAI research: FlashProve, open-source code that proves which model wrote a response, without revealing its weights. On Llama 4 Maverick: • Earlier proof: 14.3 hours per 1,000 tokens •…
ezkl is an engine for doing inference for deep learning models and other computational graphs in a zk-snark (ZKML). Use it from Python, Javascript, or the command line.
