RR trained model · Qwen

The model behind your review replies

Meet our Qwen-based model built for hospitality and review-response writing. Trained on 12 million review-related data points, RR helps your team turn guest feedback into thoughtful first drafts.

Trained on 12 million review-related data points

Specialized training for a task your team handles every day.

We trained RR for review-response writing rather than general-purpose chat. Its focus is the hospitality conversation: acknowledging guest feedback and helping your team prepare a reply in the voice of your business.

Add your property context and choose a tone, then review the result before publishing. Your team supplies the facts and makes the final call.

12M

review-related data points

Specialized training behind our review-reply model.

Tone, context, and language

What the model can be asked for through the documented RR API.

Tones the model is trained to write

Warm, Professional, and Concise are the documented tones. The free generator maps its Friendly and Warm, Professional, Concise, and Empathetic style choices onto those tones before the request is sent.

Review context the API accepts

Alongside the review text, the API documents optional context such as rating, reviewer name, manager name and title, property or hotel name, and source platform. The more context you provide, the more specific the draft can be.

Reply language

The API documents a reply language code, with English, Spanish, and French given as examples. You choose the language for the reply independently of the language the guest wrote in.

Emoji off by default

Emoji generation defaults to off. The free generator keeps it off so draft replies stay ready for professional platforms and brand review.

What that specialization is worth to a team

Less time starting from scratch. More attention to the guest.

  • Fewer blank pages. A review-response model starts from the conventions of replying to guests, so the first draft arrives close to the register managers already use.
  • Consistent voice across properties. Warm, professional, or concise can be requested per reply, which helps multi-location teams keep a recognizable tone without writing every response from scratch.
  • More languages covered. Reply language is a documented request field, so a team can answer guests in more than one language without a separate workflow.
  • Human control stays. The model drafts; a person approves. Nothing posts automatically, and the draft is always editable.

Model and provider details

Every successful RR API response reports its model and provider, plus a creation timestamp. The documented values are:

Model
qwen3-8b-review-v5
Provider
local-trained-qwen
Auth
Server-side API key

The API key stays on the server. The website generator never sends a key to the browser.

Free demo vs. paid API

Website generator — free

The free generator drafts one pasted review at a time. No account, no key, and nothing is posted.

RR API — $5 per 1,000 replies

The developer API is priced at $5.00 per 1,000 successfully generated replies, with a free plan of 100 replies per month and paid tiers above that. Pricing is configurable and may change.

See current API pricing and limits →

Drafts, not decisions

How we ask teams to use model-written replies.

  • Read and edit every draft. The model can be fluent and still be wrong, so a person approves the final text before it is published.
  • Verify every fact and commitment. Do not rely on a draft for names, dates, policies, refunds, compensation, or a specific outcome.
  • Sensitive reviews are routed to a professional tone by a simple keyword check. It can over-trigger or miss a concern, so treat the flag as a prompt to read more carefully, not as a verdict.
  • This page does not claim the model never invents details. It can, and that is exactly why human review is part of the workflow.

Questions about the RR review model

Training, tones, languages, pricing, and where the model stops.

RR is our Qwen-based model specialized for hospitality and review-response writing. We trained it on 12 million review-related data points to generate draft replies in your requested tone and language.

Put the model to work on one review

Paste a review into the free generator, pick a tone, and edit the draft it returns.