Llama · LLM monitoring
Open weights, private deployments—Meta Llama
You show up in public ChatGPT, yet your customer’s internal copilot runs on open Llama—and knows nothing about you. Meta’s Llama is the default substrate for enterprise bots and vertical assistants; it is a separate knowledge path from closed OpenAI APIs.
What a finished report looks like
The sample highlights Meta Llama’s row with quotes from Llama answers alongside closed chat models.
Carapelli
Mentions by model (demo run)
Highlight: Llama — the focus of this landing page. Numbers are illustrative.
Competitors in this slice
Your real report uses the same layout: scores, per-model breakdown, quotes, competitors, and citations — with your brand and the models you select.
Benchmarking
Timestamped snapshot
Completion time is stored with every run—clean before/after comparisons when you change positioning or content.
Method
Organic-style prompts
Your brand name is not pasted into the question text; we score whether models still mention you in realistic category queries.
Context
Around Llama
Add sibling models in the same check to see if the pattern is specific to Llama or repeats across the stack.
About this model
Llama weights ship openly; banks, retailers, and ISVs fine-tune 3.1 / 3.2 / 3.3 stacks—each host behaves differently.
There is no single “Llama experience”: RAG, adapters, and hosting vendors reshape answers, so consumer visibility ≠ corporate bot visibility.
How we measure visibility
Organic scenarios; no brand name in questions; same metrics as other Getllmspy models.
- Llama next to ChatGPT, Claude, Gemini, and more
- Share of voice, competitors, quotes
- Dated reports for internal decks
Inside the report
Snapshot header
Completion time and which models ran—your anchor for before/after benchmarking.
LLM-Score & share of voice
Aggregated 0–100 signal plus the share of models that mentioned your brand at least once.
Competitors & roundups
Who appears next to you in Llama answers: names, frequency, comparison or recommendation context.
Quotes & wording
Answer excerpts for manual review—how the model talks about the category and your brand.
Same prompts on other models
Parallel runs (Claude, Gemini, Perplexity, …) to see if the pattern is Llama-specific.
From check to PDF-ready snapshot
Brand & niche
You set brand context, site, category, language, and check type—this selects the prompt pack.
Model mix
Pick the LLM families to include; the same scenarios run in parallel across all of them.
Server run
The job executes on our side; you can close the tab and open the report from History when ready.
Report
LLM-Score, share of voice, competitors, quotes, citations—exportable and rerunnable on demand.
Big gaps vs closed chat UIs may reflect a specific host—keep rerun methodology consistent.