Granite Guardian vs Prisma AIRS AI Runtime Security API

Granite Guardian

One ungated Apache 2.0 model judges harm, jailbreaks, RAG groundedness, function-call errors and custom criteria, with signed weights and published evaluation code. It is trained and tested on English only, each call checks one criterion, the 4.1 prompt format differs from 3.x, and IBM's watsonx.ai lists only the deprecated 3.0 model.

Prisma AIRS AI Runtime Security API

A public OpenAPI document, ten detection types in one call, tool-call scanning, a remote MCP server, rotating API keys and OAuth roles suit teams already on Strata Cloud Manager. Access needs Software NGFW credits bought through sales, with no public price, trial or self-serve signup, and payloads flagged malicious are kept for up to 10 years.

Granite Guardian

1. Append the <guardian> block as the final user message, with the mode line, ### Criteria: and ### Scoring Schema:. Copy the strings from the model card, because no package builds them
2. Use the no-think instruction for gating and parse <score>. Think mode writes a reasoning trace first, and the card's examples allow up to 2,048 output tokens
3. Treat yes as the criterion being met, which for built-in criteria means the risk is present. Treat a missing <score> tag as a failed check
4. Pass retrieved text through documents= and tool schemas through available_tools= in apply_chat_template, not inside the message text
5. Under Ollama, set num_ctx in the request options. IBM's docs say the default context is short and long requests are truncated

Prisma AIRS AI Runtime Security API

1. Send x-pan-token and an ai_profile with profile_name or profile_id on every scan. Both come from Strata Cloud Manager, and a key works only in its own region
2. Put earlier turns first in contents. Only the last element is scanned, and the rest is context
3. Read action (allow or block) and category, then check timeout and error, because a detector that failed is reported in errors with a 200
4. Keep synchronous requests under 2 MB and asynchronous ones under 5 MB and 25 items, and fetch at most 5 scan or report IDs a call, 10 calls a minute
5. For MCP, connect to /mcp on the regional host with x-pan-token and x-pan-profile headers, and call pan_inline_scan yourself before and after generation