SHERIDAN, WY — May 12, 2026 — Safeguard today announced the general availability of Griffin Zero, the flagship 671-billion-parameter mixture-of-experts reasoning model trained exclusively on cybersecurity data, for customers deploying in the company's Sovereign and Air-Gapped tiers. Griffin Zero is the largest model in the Safeguard lineup and is purpose-built for cross-repository taint analysis, candidate zero-day reasoning, and regulator-grade evidence generation.
Griffin Zero supports a 256,000-token context window, allowing the model to hold an entire mid-sized service graph — including dependency manifests, call paths, and configuration artifacts — within a single reasoning pass. Internal benchmarks measured a 12 percent precision uplift on cross-package taint paths versus the prior generation, with the largest gains observed on long-chain transitive flows that pattern scanners typically miss.
Deployment sizing scales from an entry configuration of 11 NVIDIA H100 accelerators up to a 22 H100 configuration optimized for tenant-isolated regulator workloads. Both configurations support fully offline operation with no outbound network dependency, which is a baseline requirement for FedRAMP HIGH-ready and STQC-ready environments. Customers can pin model weights, retrieval indexes, and policy bundles to an internal artifact store and operate the platform without contacting any Safeguard-hosted service.
"Griffin Zero exists because regulator-grade software supply chain reasoning cannot be outsourced to a general-purpose chatbot — it has to be done by a model that has only ever read security artifacts, and it has to run where the regulator says it has to run," said Hritik Kumar Sharma, Founder and CEO of Safeguard. "Today's release is the version of Griffin Zero we have been validating with sovereign and air-gapped customers, now available for general deployment."
Beyond raw model capability, the general availability release includes signed model attestations, reproducible weight hashes, and a runtime verifier that ensures the deployed binary matches the published manifest. These artifacts are intended to satisfy supply chain integrity requirements that increasingly apply to the AI components themselves, not only the software they analyze.
"The most useful thing Griffin Zero does is shorten the distance between a finding and a citation," said the Safeguard engineering team. "Reviewers can see the call path, the sink, and the evidence chain in one place — without leaving the air gap."
Griffin Zero is available immediately to customers in the Sovereign and Air-Gapped tiers. Smaller members of the Griffin family — Griffin L, M, S, and Lite — remain available for teams whose deployment envelopes do not require the full 671B model.
About Safeguard
Safeguard is the software supply chain security platform that fuses 11 scanners, a security-only AI model lineup (Griffin · Eagle · Lino), and reachability-aware reasoning to find what pattern scanners miss — from CVEs to candidate zero-days — and to ship the fix with cited reasoning. The platform is built for engineering teams shipping production software and for the regulators auditing them. Learn more at https://safeguard.sh.
Media Contact
press@safeguard.sh