No model card yet.
Build a request for r_debab/security-one. The form and the JSON are the same object; the response panel shows the shape every System One server returns.
Placeholder values, in the shape a run returns.
safenoul7%This model doesn't answer live here. Send this request to Superagent's endpoint or your own server for this model; every System One server speaks this shape.
curl https://api.systemonemodels.ai/v1/systemone \
-H "Authorization: Bearer $SYSTEMONE_TOKEN" \
-H "Content-Type: application/json" \
-d '{"model":"r_debab/security-one","state":"Task: clean up inactive accounts before the quarterly report.\nProposed tool call: delete_rows(table=\"customers\", where=\"last_login < 2023-01-01\")\nContext: the customers table is production; there is no backup newer than Friday.","questions":{"safe":{"type":"noul","instructions":"Is this action safe to run without a human approving it first?"}}}'Where this model runs today. Call the provider directly or run the weights yourself.
| Provider | Status | Input | Output | Context | Link |
|---|---|---|---|---|---|
| Superagent | Generally available | — | — | 256K | Get access |
| Your own hardware | Open weights | Free | Free | 256K | Install from its source repository |
Prices are the provider's published rates per million tokens; get a key from Superagent.
| Decision accuracy | 99.8% |
|---|
Suite: BIPIA binary prompt-injection detector adaptation (800 rows; Superagent's frozen evaluation, unsafe if P(unsafe) >= 0.70). Numbers are the publisher's own, from the manifest; the model card says how they were measured.
No versions published yet.
Every ancestor down to this model, then what was built on it, wherever each one lives.
Models marked ↗ are on Hugging Face and not on the registry yet.
{
"spec_version": "0.1",
"model": "security-one",
"namespace": "r_debab",
"category": "system-one",
"architecture": "security-one",
"base_model": "denis-pplx/autojev-27b",
"base_model_source": "huggingface",
"license": "apache-2.0",
"license_name": null,
"summary": "Superagent's 27B security decision model. It scores 2 to 16 supplied options in one forward pass and returns a probability for each, for triaging prompt injection, agent tool calls, code changes, logs and alerts. Apache-2.0 weights (gated, auto-approved) plus a hosted API.",
"tags": [
"system-one",
"security",
"prompt-injection",
"qwen",
"sglang",
"hosted",
"gated",
"27b"
],
"capabilities": [
"choice",
"score",
"noul",
"classify",
"route"
],
"runtime": {
"framework": "sglang",
"entrypoint": null,
"python": null,
"hardware": "gpu"
},
"artifacts": [],
"evaluation": {
"suite": "BIPIA binary prompt-injection detector adaptation (800 rows; Superagent's frozen evaluation, unsafe if P(unsafe) >= 0.70)",
"suite_version": null,
"decision_accuracy": 0.9975,
"calibration_error": null,
"valid_action_rate": null,
"median_latency_ms": null,
"p95_latency_ms": null
},
"parameters": 27000000000,
"context_length": 262144,
"hub_repo": "superagent-ai/security-one-27b",
"released": "2026-09-25",
"availability": "both",
"hosted": {
"provider": "Superagent",
"url": "https://api.superagent.sh",
"docs_url": "https://www.superagent.sh/docs/models/api",
"input_price_usd_per_mtok": null,
"output_price_usd_per_mtok": null,
"status": "generally-available"
}
}View raw metadata It answers 5 kinds of question: choice picks one option from a list you give it; score rates the state on a scale you define; noul returns the probability that a condition holds; classify assigns the state to one or more labels; route sends the state to one of several handlers. Every answer is a typed value with a probability, never free text.
The registry lists it without files: install it from the source repository linked on this page. Superagent serves it over an API; the Providers section has the prices and the link.
It is published under apache-2.0. Read the licence before shipping; the model card may add terms of its own.
Its base model is denis-pplx/autojev-27b, a security-one architecture. The Model tree section shows the whole chain.
Up to 262,144 tokens of state per request. The model has 27B parameters.
Its publisher reports 99.8% decision accuracy on BIPIA binary prompt-injection detector adaptation (800 rows; Superagent's frozen evaluation, unsafe if P(unsafe) >= 0.70). Suites differ between publishers, so compare models on your own labelled examples.
This page is https://systemonemodels.ai/r_debab/security-one. Its metadata is at https://api.systemonemodels.ai/v1/models/r_debab/security-one and a Markdown summary for agents at https://systemonemodels.ai/r_debab/security-one.md.