Model Family · Weighted for Cybersecurity

Three models. One brain. Trained only on what attackers actually do.

A single general-purpose model is the wrong tradeoff curve for security work — latency, precision and cost pull in different directions, and one set of weights can't sit at all three optima. The lineup splits the work: Lino runs inline at sub-100 ms, Eagle sweeps the repo and ranks candidates, Griffin spends a real reasoning budget proving and disproving each one.

Why three models

One curve doesn't fit precision, latency, and cost.

Each model occupies a different point on the security tradeoff curve. You don't pay for reasoning you don't need and you don't wait for latency you can't spend.

Sub-100 ms or it's not inline

A model that takes two seconds to think is a model the developer disables. Lino is sized to answer before the editor finishes redrawing — every other choice flows from that constraint.

Wide angles need ranking, not reasoning

Sweeping a repo for taint candidates is a clustering problem, not a chain-of-thought problem. Eagle's head is fine-tuned for ranking dataflow tokens — fast, batched, and cheap enough to run across every package on every commit.

Precision requires reasoning budget

Hypothesising an exploit chain and trying to refute it under sanitiser constraints is genuinely hard. Griffin gets the parameter count, the context window, and the chain-of-thought depth to earn its verdict.

The lineup

Griffin · Eagle · Lino — one corpus, three speeds.

Griffin

01

Deep reasoning · the hypothesis engine.

Where it runs:Cloud-only · heavy reasoning budget · per-tenant inference isolation.
Architecture
  • ~70B mixture-of-experts head
  • Security-augmented tokeniser (CWE-IDs, CVE-IDs, taint-source tags, dataflow operators)
  • 128k context window for multi-hop call graphs
  • Multi-step chain-of-thought with adversarial disproof pass
  • Distillation source for both Eagle and Lino
Training corpus
  • Curated CVE bodies and accompanying patch diffs
  • Exploit write-ups and public proof-of-concept code
  • Vulnerability disclosure threads and security advisory text
  • Malware behavioural traces and reverse-engineering notes
  • IDA / Ghidra disassembly excerpts with annotations
  • Cross-package taint graphs extracted from open-source repos
  • MITRE ATT&CK mappings, security RFCs, package registry metadata
Eval highlights
81%
Exploit-hypothesis accuracy
98%
Adversarial prompt resistance
0.6%
Hallucination rate on security Q&A
128k
Reasoning context window
Size variants

Griffin scales smoothly from an 8B distilled head to a 671B-parameter mixture-of-experts. Customers pick the variant whose latency budget and reasoning depth match the workflow — and Safeguard can route between variants automatically based on finding complexity.

VariantParametersContext windowLatency p95Deployment shapeTypical use
Griffin Lite8B32k~1.2sIDE-side cloud burst / CLI deep-scanFast remediation suggestions on a single finding.
Griffin S14B64k~2.8sCloudMid-depth call-graph reasoning, PR-level reviews.
Griffin M32B128k~5.5sCloudRepo-wide reasoning, transitive taint chains.
Griffin L70B128k~8sDedicated GPUMulti-hop cross-package exploit hypothesis. Default Safeguard production tier.
Griffin Zero671B-MoE (sparse, ~37B active)256k~12sMulti-GPU cluster / sovereignDeepest reasoning, supply-chain-scale audits, agentic disclosure workflows.

Latencies are p95 on representative single-finding workloads. Context windows are usable, not theoretical maxima.

Eagle

02

Wide-angle triage · the surface scanner.

Where it runs:CI + cloud · batched across the whole repo on every push.
Architecture
  • ~13B dense transformer, distilled from Griffin
  • Ranking + clustering head fine-tuned for taint-path prioritisation
  • Attention biased toward dataflow tokens (sources, sinks, sanitiser ops)
  • Batched inference optimised for full-repo sweeps
  • Same security-augmented tokeniser as Griffin
Training corpus
  • Taint graphs and call-graph slices labelled for reachability
  • SAST findings with ground-truth exploitability labels
  • Fuzzing corpora and crash-derived input grammars
  • Sanitiser configurations from real-world projects
  • Package registry metadata with version-pinned advisories
  • CWE-class labelled patch sets
Eval highlights
94%
Top-5 taint path recall
87%
Triage precision
420 ms
p95 latency per package
13B
Parameters, dense

Lino

03

Inline · the commit-time gut check.

Where it runs:IDE / CLI / pre-commit · 100% on-device, no source code leaves the machine.
Architecture
  • ~1B parameters, distilled from Griffin via security-task distillation
  • INT8 quantised weights, runs locally with no network egress
  • Optimised for sub-100 ms inline latency on a developer laptop
  • Same security-augmented tokeniser; pruned to inline-relevant vocab
  • Ships in the IDE extension, CLI, and pre-commit hook
Training corpus
  • Distillation traces from Griffin on common sink + sanitiser pairs
  • Hand-labelled inline patterns for unsafe deserialisation, command injection, SSRF, path traversal, prompt injection
  • Sanitiser-aware negative examples to suppress noise
  • Public-only training; no customer code, ever
Eval highlights
80 ms
p95 inline latency
0.79
F1 on inline sink detection
100%
On-device, no egress
1B
Distilled parameters
How the variants route

Complexity picks the parameter count.

Every finding gets a complexity score; the score chooses the Griffin variant. You do not pay for 671B reasoning on a one-hop sink, and you do not waste a 70B verdict on a multi-hop chain you actually need to prove.

01
Triage score

Eagle assigns each candidate a complexity score from the call graph: depth, sanitiser ambiguity, cross-package edges, sink severity.

02
Variant selection

Cheap, in-package candidates route to Griffin Lite. Mid-depth PR work routes to S or M. Multi-hop cross-package paths route to L. Sovereign or classified workloads get Zero.

03
Reasoning pass

The chosen variant runs the hypothesise → cite → disprove → patch trace. The trace ships with the finding so reviewers can audit which variant produced what.

Cheap candidates go to Lite. Multi-hop cross-package paths route to M or L. Sovereign and classified deployments with the longest budget reach Zero.

Inside the Griffin head

Architecture details, not marketing copy.

Mixture-of-experts routing in Zero

Zero uses 8 experts with top-2 routing per token, activating roughly 5.5% of the 671B parameter count per forward pass. Routing is learned, security-task-aware, and biased toward sink-handling experts when the input embedding looks like a taint sink.

Security-augmented tokeniser

Around 28k extra tokens cover CWE / CVE identifiers, taint operators, package coordinates, and attack-pattern shorthand. Names like preg_replace, ObjectInputStream and pkg:npm/lodash do not shatter into byte-pair noise — their embedding-space neighbours are other vulnerable patterns.

Long context, retrieval-augmented

Sliding-window attention plus landmark tokens gives a 256k usable window on Zero. The call-graph chunker retrieves the slice of the program that matters for a given finding and pages it in around the hypothesis, so context is full but coherent.

Structured reasoning trace

Every Griffin call emits a four-stage trace: hypothesise the exploit, cite the path in the call graph, propose a disproof, propose a patch. The trace is the finding — reviewers see the chain, not a single label, and can reject at any stage.

Security-domain RLHF

Preference data is labelled by senior offensive-security engineers, not generic annotation vendors. The reward model penalises plausible-sounding hallucinations on CWE classification and treats unverified reachability claims as failures.

Distillation source for the lineup

Griffin (any size) is the teacher for Eagle and Lino. Their behaviour stays consistent with Griffin's reasoning on the same finding — they are smaller, faster heads of the same brain, not a different model trained to imitate.

Deployment shapes

Where each tier actually runs.

The lineup is the same brain at four isolation levels. Pick the shape that matches your data-residency and trust boundary; the variants stay consistent.

Shared cloud

Eagle + Griffin Lite / S

Multi-tenant inference, fastest onboarding, per-tenant isolation at the prompt and KV cache layer. Best for teams trialling the lineup.

Dedicated cluster

Eagle + Griffin S / M / L

Single-tenant inference on isolated VPC hardware. No cross-tenant traffic, deterministic latency, full SHA-pinned weight attestation.

VPC-isolated

Eagle + Griffin L

Customer-controlled VPC, bring-your-own-key encryption, no cross-tenant traffic. Inference plane sits inside your network perimeter.

Sovereign / air-gapped

Full lineup including Griffin Zero

STQC-ready, no internet egress, on-prem GPU, full audit-log export. Suits regulated, classified, and sovereign deployments.

How they hand off

Commit to queue, one pipeline.

01Lino
Commit

Inline sink + sanitiser check on the developer's machine.

02Eagle
Repo scan

Cross-package taint paths ranked across the codebase.

03Griffin
Path hypothesis

Deep reasoning posits an exploit chain + CWE class.

04Griffin
Adversarial disproof

Second pass tries to refute; survivors land in your queue.

Every survivor lands in your queue with the taint path, hypothesised trigger, disproof attempt log, and a ranked evidence bundle.

Cybersecurity-only training

Why the corpus matters more than the parameter count.

Refusal rates drop when the corpus knows you're a defender.

General-purpose models are trained to flinch at the word 'exploit'. Ours are weighted on disclosures, write-ups, and patch diffs — the literature of defenders — so a question about a deserialisation gadget gets answered, not refused.

The tokeniser speaks security.

CWE-IDs, CVE-IDs, taint-source tags, and dataflow operators are first-class tokens, not stochastic fragments. Sink names like preg_replace and ObjectInputStream don't shatter into byte-pair noise. Embedding-space neighbours are other vulnerable patterns, not unrelated English.

No contamination from marketing copy or unframed code.

General code corpora are dense with patterns that look secure to a model trained on README files and tutorials but ship CVEs in production. We deliberately exclude product copywriting, StackOverflow snippets without a security frame, and any code lifted without disclosure context.

RLHF rubric written by security engineers.

Reward modelling is graded against a rubric that penalises plausible-sounding hallucinations on CWE classification, rewards explicit reachability reasoning, and treats unverified claims as failures. The rubric is owned by the security team, not a generic annotation vendor.

Closing the loop with the engine.

Every finding the deterministic taint engine surfaces becomes a structured brief the model sees in inference. The model doesn't have to guess at reachability — it inherits it. That hybrid is what keeps the false-positive rate honest.

What's NOT in the corpus

The exclusion list is contractual, not aspirational. Every item below is audited at ingest.

  • No customer code — ever, at any tier.
  • No general web crawl or social media text.
  • No PII, secrets, or scraped credentials.
  • No closed-source proprietary disassembly.
  • No LLM-generated synthetic security text.
  • No product copywriting or vendor marketing.
  • No StackOverflow snippets without a security frame.

What IS in the corpus

Defender-frame data, deduplicated and labelled. The ingest manifest is reviewable under NDA.

  • CVE descriptions plus patches with linked diffs.
  • Exploit write-ups and POCs, curated and deduplicated.
  • Vendor advisory texts (CERT, MITRE, vendor PSIRT).
  • Static-analysis findings with ground-truth labels.
  • Taint graphs extracted from open-source repos.
  • MITRE ATT&CK technique descriptions and procedure examples.
  • Fuzzing corpora and crash triage notes.
  • Package-registry metadata and malware behavioural traces.
  • IDA / Ghidra disassembly excerpts, security-tagged.
  • Vulnerability disclosure threads, with consent.
  • Security RFCs and standards (NIST, ENISA, OWASP).
Provenance & safety

Guarantees in writing.

Tenant-level commitments

  • Customer code never enters training — contractual, audited, and per-tenant attested.
  • Per-tenant inference isolation; no cross-tenant prompt or KV cache reuse.
  • Full model + dataset attestation available under NDA, including SHA-pinned weights.
  • Opt-out for any analytics signal contributing to evaluation telemetry.
  • Export-controlled deployments available for federal customers (FedRAMP HIGH / IL5 path).
  • Coordinated disclosure with upstream maintainers stays opt-in per finding.
Benchmarks at a glance

Numbers next to the alternatives.

Same internal eval suite, run against a leading general-purpose model and a representative pure-LLM bug hunter. Lower is better for FPR, latency, and refusal; higher is better for resistance and recall.

StackFalse positive rateAdversarial prompt resistanceCross-package taint path recallLatency p95 (inline)Refusal rate on security Q&A
General-purpose LLM (zero-shot)62%41%18%1.8 s34%
Pure-LLM bug hunter (Mythos-class)54%63%37%2.4 s11%
Safeguard Griffin Lite (8B)19%91%78%1.2 s3%
Safeguard Lineup (Lino + Eagle + Griffin)6%98%94%80 ms<1%
Safeguard Griffin L (70B)4%98%96%8.0 s<1%
Safeguard Griffin Zero (671B-MoE)2%99%98%12.0 s<0.5%
Numbers from Safeguard's internal eval suite. Methodology available under NDA.

Run the lineup against your code.

Lino at your commit, Eagle across your repo, Griffin proving the survivors. See the candidates your scanner never surfaced.