GET
/v1/models/ollaya/decima:guard
Accept:
| rawId | decima:guard |
| displayName | — |
| id | ollaya-decima-guard |
| activity | decision |
| schemaEndpointId | no bound generation route |
| releasedAt | 1792281600 — Oct 18, 2026 |
| firstSeenAt | 1792281600 — Oct 18, 2026 |
| lastSeenAt | 1795358012 — 3m ago |
| pricing | not published by this provider’s list endpoint |
| contextWindow | 2048 |
| openWeights | true |
| weightsUrl | "https://huggingface.co/amyrmahdy/decima-guard" |
| providerMetadata | {"model_format":"onnx","family":"decima","parameter_size":"321M","context_length":2048,"languages":["multilingual"],"description":"A. M. Madani's Decima-guard 1.0 preview: a guard for LLM apps on Decima-base-clean (mmBERT-base) that checks user messages, tool results and model replies for prompt injection, jailbreaks, leaks and harmful content (the OWASP Top 10 for LLM applications), trained for English, Arabic and Persian. 321M parameters. States up to 2,048 tokens.","source":"huggingface.co/amyrmahdy/decima-guard@98d3f734c8323cb390d9ad750e1a606a8dd56f09","license":"Apache-2.0","release_date":"2026-10-18"} |
| maxOutput · modalities · capabilities · knowledgeCutoff · requestMap | not published by this provider’s list endpoint |
Endpoints & schemas
GET /v1/schemas/ollaya →| endpoint | activity |
|---|---|
| v1/systemone | decision |
Use it
shell
$ curl https://modelschemas.com/v1/schemas/ollaya/decision/v1/systemone # request schema
$ curl -X POST https://modelschemas.com/v1/validate -d '{"provider":"ollaya","endpointId":"v1/systemone","payload":{…}}'
$ curl -X POST https://modelschemas.com/v1/estimate -d '{"provider":"ollaya","model":"decima:guard","usage":{"input_tokens":1200,"output_tokens":400}}'