We can finally share this! We’re bringing Bitcoin Red Team together with @LuganoPlanB on 22 October, right here in our home city. Super excited & looking forward to getting Bitcoin developers and security researchers in the same room to talk through how AI can help with security reviews, what still needs expert judgment, and what we can learn from recent incidents. Joining us are @Rob1Ham , @danielabrozzoni, @KLoaec, @MrKukks and @giacomozucco , alongside Federico N. and @suhanprabhu from Prem AI. If you build, maintain or review Bitcoin software, we’d love to have you there. Bring your questions and experiences. 22 October · 15:30–18:30 CEST @pow_space, Lugano · Sessions in English Request an invitation through the post below. Space is limited, and the hosts will review requests and confirm places.

Prem Confidential API's and ZDR API's have already processed 3.8 billion tokens across 124,000 requests and 11 models. None of those 3.8 billion tokens is sitting on a server anywhere. Your prompts and completions aren't written to a log or a cache, and they're not used to train anything, so whatever you're working on stays yours. If you've been wanting to try a strong model but didn't love the idea of your sensitive data living on someone else's server, ZDR is for you.

When a Bitcoin security team asked @premai for API access to Kimi K3 the week it dropped, we saw the problem most builders quietly live with: every time a new frontier model ships, teams scramble for access, rebuild an integration, and hand their prompts to yet another provider that keeps them. Then the same request kept arriving from other teams, a different model each time, and the opportunity became obvious: one endpoint that carries every new frontier release the day it lands, speaks the OpenAI format your client already uses, and stores nothing. So Prem built it, and today ZDR is live as part of Prem API: zero data retention across Kimi K3, GLM 5.3 Flash, Qwen 3.8, and DeepSeek V4 behind one OpenAI-compatible base URL, where neither Prem nor its inference partners retain prompts or completions, and where the trust model is stated as precisely as the feature set, since ZDR runs on contracts and operational controls with plaintext requests at the gateway for evals, prototypes, and public data, while Prem Confidential Inference covers source code and sensitive production traffic with client-side encryption and attestation. If your team wants day-one frontier access without retention, one API with two explicit trust models is waiting, and the comparison will help you pick the right one per workload: docs.prem.io/zdr/comparison

Remember ox-alpha? The anonymous model became the most popular model of the week during testing on OpenCode and OpenRouter. It was GLM 5.3 Flash: 320B total parameters, only 18B active, and remarkably strong coding and agentic performance. GLM 5.3 Flash is now available in Cyberscan as the Balanced option.

Some long Cyberscan reviews were stopping near the end and making users repeat completed checks. We fixed the recovery path. Accepted checks now stay with the review. If a scan is interrupted, Continue resumes finding verification from the first unfinished verdict. Completed verification is saved too, and you pay only for new work after you continue. We also added static security review for first-party PHP code. This extends Cyberscan to more payment, commerce and Bitcoin infrastructure repositories, while keeping dependency, generated and test paths out of the review. Also - we have also issued refund credits for any incomplete runs/outcomes. We thank you for all the testing and feedback so far. Keep it coming!

Read the full changelog and run a scan: cyberscan.prem.io/changelog