Safeguard
Model Family · Capabilities

Every task. Which model handles it.

The comprehensive task-to-model mapping for the lineup. Every capability the family ships, mapped to the variant that actually performs it, with a per-cell verdict — full support, partial support, not supported, or dispatched to a sibling via the auto-router.

Lineup recap

Three families, one corpus.

Lino

Inline · sub-100 ms · on-device · the commit-time gut check.

Read the deep dive

Eagle

Wide-angle triage · ranks and clusters candidate paths across the repo.

Read the deep dive

Griffin

Deep reasoning · the hypothesis engine, five variants from 8B to 671B-MoE.

Read the deep dive
Master capability matrix

Capability by variant.

Forty-two tasks across seven bands. Marks reflect actual capability rather than badge-collection upgrades.

Full support
Partial
Not supported
Via routing
Capability
Lino
1B
Eagle
13B
Griffin Lite
8B
Griffin S
14B
Griffin M
32B
Griffin L
70B
Griffin Zero
671B-MoE
Detection & triage
Inline sink detection (deserialisation, SSRF-able URL builders, unsafe SQL)
Secret pattern detection (Gitleaks-class)
Sanitiser-quality scoring
Cross-scanner finding dedup
Taint path enumeration (single-package)
Taint path enumeration (cross-package)
Path ranking + clustering
Confidence scoring per candidate path
Reasoning & hypothesis
Exploit-class hypothesis (CWE category mapping)
Exploit-trigger input synthesis
Cross-package taint chain reasoning (≤ 4 hops)
Cross-package taint chain reasoning (≤ 12 hops)
Cross-package taint chain reasoning (> 12 hops)via routing
Multi-finding correlation in a single reasoning pass
Adversarial disproof pass (refute own hypothesis)
Structured reasoning trace output
Remediation
Single-finding fix suggestion
Auto-fix PR with diff
Auto-fix PR with cited reasoning trace
Sanitiser-aware patch synthesis
Multi-service auto-fix campaign
Upstream coordinated-disclosure patch + draft
Eval & gates
PR-time gate decisionvia routing
Pre-merge policy evaluationvia routing
SARIF / CycloneDX / SPDX emit
Eval-harness scoring of candidate patches
Context & scale
Context window 32k
Context window 64k
Context window 128k
Context window 256k (retrieval-gated)
Repo-wide reasoning (1k–5k packages)
Portfolio-wide reasoning (multi-repo)
Deployment shape
On-device inference (no network egress)
Shared cloud
Dedicated cluster
VPC-isolated
Air-gapped / sovereign
AI & MCP governance
Sensitive-data egress scanning
Prompt audit-log signing
MCP tool-call inspectionvia routing

"Via routing" means the variant does not perform the capability itself; the auto-router dispatches to a sibling when that capability is requested.

Per-model quick reference

One card per variant.

01

Lino 1B

Inline gut check at the keystroke.

Best at
  • Sub-100 ms inline sink + sanitiser detection
  • Secret patterns and obvious unsafe primitives
  • On-device, fully offline operation
In the pipeline: Sits in the IDE, CLI, and pre-commit hook.
Deep dive
02

Eagle 13B

Wide-angle triage across the repo.

Best at
  • Ranking and clustering candidate taint paths
  • Cross-scanner deduplication and confidence scoring
  • Batched full-repo sweeps after CI
In the pipeline: Runs after CI; feeds the auto-router queue.
Deep dive
03

Griffin Lite 8B

Cheapest deep reasoning tier.

Best at
  • Single-finding reasoning at low latency
  • Sanitiser-aware fix synthesis
  • Short-hop taint chain proofs
In the pipeline: Low-complexity candidates from Eagle's queue.
Deep dive
04

Griffin S 14B

PR-level reasoning, 64k window.

Best at
  • Mid-depth call-graph reasoning
  • Adversarial disproof on PR diffs
  • Structured reasoning trace output
In the pipeline: PR reviewer tier; default for mid-confidence findings.
Deep dive
05

Griffin M 32B

Repo-wide reasoning, 128k window.

Best at
  • Cross-package taint chains up to twelve hops
  • Multi-finding correlation in one pass
  • Repo-scale (1k–5k package) reasoning
In the pipeline: Medium-confidence findings, transitive chains.
Deep dive
06

Griffin L 70B

Default production tier.

Best at
  • Multi-hop cross-package exploit hypothesis
  • Cited auto-fix PRs with full trace
  • Portfolio-wide reasoning across repos
In the pipeline: High-confidence findings; the survivors-of-disproof tier.
Deep dive
07

Griffin Zero 671B-MoE

Deepest reasoning, sovereign tier.

Best at
  • 256k retrieval-gated context
  • Coordinated-disclosure draft synthesis
  • Supply-chain-scale audits
In the pipeline: Sovereign and air-gapped deployments only.
Deep dive
How the auto-router decides

Triage score picks the variant.

The router is not a vibe. A single scalar collapses confidence, complexity, hop depth, and tenant tier into a bucket that maps to one variant.

01
Lino runs at the commit

Inline on the developer machine, sub-100 ms, no network egress. Catches the obvious unsafe primitives before they ever reach a build server.

02
Eagle sweeps the repo

Post-CI batched scan across every package. Ranks and clusters candidate paths and assigns each one a confidence score on the dataflow head.

03
Auto-router scores triage

The ranked queue is scored on a 0.0–1.0 scale using confidence, complexity, hop depth, and the deployment tier the tenant is on.

04
Triage score picks the variant

0.0–0.4 → Lite · 0.4–0.6 → S · 0.6–0.75 → M · 0.75–0.9 → L · 0.9–1.0 → Zero (sovereign tier only). The chosen variant emits the reasoning trace.

Triage score → Griffin variant
0.00–0.40
Griffin Lite
0.40–0.60
Griffin S
0.60–0.75
Griffin M
0.75–0.90
Griffin L
0.90–1.00
Griffin Zero

Zero is only reachable for tenants on a sovereign or air-gapped deployment tier. Below 0.4, Eagle's verdict ships as-is and no Griffin pass is requested.

Honest about scope

What's NOT on this matrix.

The lineup is weighted for cybersecurity. It deliberately doesn't try to do the things below — a different model class would be the right tool.

Out of scope by design

  • General-purpose code generation outside a security context.
  • Image, audio, or video generation of any kind.
  • Customer-support or open-domain chat.
  • Translation, summarisation of marketing copy, or content rewriting.
  • Autocomplete for unrelated business logic.
  • Code review for style, lint, or developer-experience concerns.
  • Synthesising training data for unrelated downstream models.
  • Hardware fault diagnosis or non-software risk analysis.

The corpus is curated to defenders, taint graphs, CVE bodies, and patch diffs. We'd rather be excellent at a small set of security tasks than mediocre at a large set of unrelated ones.

Pick the right variant for your workload.

The matrix is the spec sheet. Book a call and we'll map your workflow onto the variants — including which findings the router should escalate, and which it should let Eagle settle.

Lineup overviewResearch notes