Standalone model
IntroducingAlpha · In Development
Quantum Sentinel-Alpha
Specialized AI for secure software.
A specialized AI model being trained to find, understand and fix vulnerabilities in source code.
// Planned report format
finding {
- location
- <file>:<line range>
- cwe
- CWE-<id>// when evidence is sufficient
- evidence
- data flow · trust boundary
- remediation
- concrete mitigation
- patch
- minimal, security-focused// optional
- verification
- tests · re-analysis
}
Evidence-backed findingRejectedAbstained
- Status
- Alpha · In Development
- Alpha target
- October 2026
- Primary foundation candidate
- Qwen3.5-9BFinal selection after the internal 4B / 9B security bake-off.
- Primary languages
- Python · JavaScript / TypeScript · Go
- Cybersecurity-focused
- Local-first
- Open weights planned
Planned architecture
A specialized model. A verifiable system.
The model evaluates semantic risks. The Sentinel Engine supplies repository context and deterministic tools, then checks candidate findings. The planned model remains independently downloadable and locally usable.
Repository / Pull Request
Sentinel EngineRepository Indexer
Sentinel EngineStatic Signals + Context
Sentinel EngineQuantum Sentinel
Quantum SentinelCandidate Findings
Quantum SentinelVerifier / Tests / Re-analysis
Sentinel EngineEvidence-backed Finding
Sentinel EngineJSON / SARIF / Human Report
Sentinel Engine
Context + deterministic tools
Sentinel Engine
Deterministic work belongs in tools: indexing narrows the search, static analysis adds signals, verification checks model claims and tests help establish whether a patch still works and is safer. Specialization plus verification matters more than model size alone.
Security training objectives
From a potential risk to a reasoned finding.
Training objectives for Alpha. Performance has not yet been published. When evidence is insufficient, the model should express uncertainty or abstain.
Detection
Find potential vulnerabilities in source code while distinguishing unsafe patterns from benign lookalikes.
Localization
Identify affected files and relevant code spans, rather than returning only generic warnings.
CWE Classification
Map findings to appropriate CWE categories when sufficient evidence exists.
Evidence
Provide concise technical evidence around data flow, trust boundaries and why a pattern may be unsafe.
Remediation
Explain concrete ways to remove or mitigate a vulnerability.
Patch Generation
Propose minimal security-focused patches designed to preserve existing functionality.
False-Positive Rejection
Recognize similar-looking, safe code as benign. Abstain instead of inventing a vulnerability.
Secure Review
Review repositories, pull requests and diffs as a specialized second security reviewer.
Development plan
Alpha first. A stable release later.
Alpha is the first real security-SFT milestone, before Beta, Release Candidate and the later stable Quantum Sentinel 1.0. The early Lean plan aims for an initial Alpha fine-tune; the sequence below is a development plan, not a record of completed phases.
Alpha target
October 2026
An Alpha development target, not a guaranteed release date. Quantum Sentinel 1.0 follows later.
Foundation
Planned
Baseline
Planned
Dataset / Garden
Planned
Calibration
Planned
SFT Alpha
Current focus
Data Iteration + Beta
Planned
Optional Hardening
Planned
Release Candidate
Planned
Quantum Sentinel 1.0
Planned
rappidAI Research
Independent AI research. Berlin.
We build compact language models and open tools for inspectable AI systems. Independent, founder-led research from Berlin, with Quantum Sentinel-Alpha as our current security-model focus.
Meet rappidAIThe work
Intelligence. Under your control.
From model weights to agent execution. Three distinct research directions, built around efficiency, transparency and local control.
rappidAI Quantum
Compact models. Open experiments.
Security-specialized Sentinel, the Echelon model line and German-language pilot models. Explore the releases, methods and evidence behind each stage.
Sentinel Alpha + Echelon in development · Two public pilots
Explore our modelsrappidAI Ghost
Set the boundaries.
An integrated security runtime for autonomous agents: isolation, SHADOW deception, prompt-injection signals, scoped approvals, runtime limits and evidence behind one simple flow.
v0.3.0 · Integrated agent security · Experimental
ExplorerappidAI Replay
Understand every run.
Record execution, restore technical state and compare branches. Local-first infrastructure for reproducible agent research.
Experimental · Open source
ExploreOur approach
Built to be understood.
Useful research leaves a trail. We connect our work to code, model artifacts and documented methods, so you can examine what works and where the limits are.
Inside the research- Efficiency
- Compact architectures and local inference, with compute treated as a real constraint.
- Control
- Inspectable systems and explicit boundaries around what an agent can access.
- Evidence
- Published results, current development and open questions are clearly distinguished.
From the lab
Research you can inspect.
rappidAI
Independent. Focused. Berlin.
rappidAI is an early-stage, self-funded research and development initiative building compact intelligence and the infrastructure around it.