Multiple subscription tiers and a stronger-model premium plan

Category: Feature request Priority: P3 Area: Product / monetization Reports: 2 threads Status: Open

Summary

Two users independently propose a monetization structure for Qaf. Majed Al-Harbi guesses that the free, ad-free app must be running an older API model and suggests a paid subscription that unlocks a stronger LLM (he names "GPT-5.6 Sol") to improve the accuracy of quotations and answers — while stressing he has not actually noticed any inaccuracy yet. Mahmoud bin Ahmed Al-Harouni, a blind Android screen-reader user who sent a long, structured feedback letter, argues full free access will be hard to sustain given inference/infrastructure costs and proposes several tiers (Premium / mid / Lite) plus a preserved free plan with usage limits, so students and researchers are not priced out. Al-Harouni also suggests evolving Qaf from a research assistant into a full Islamic assistant (prayer times by location, other practical services) and integrating with Quran.com APIs. Abdellatif engaged Majed on pricing (asked what monthly amount seems fair; Majed deferred, suggesting they model the costs with AI).

What users say

ماجد الحربي (Majed Al-Harbi) — ma2004jed3@gmail.com

• Date: 2026-08-15
• Language: Arabic
• Thread: 1a006f06be307f8e (subject: "اقتراح إمكانية ترقية النموذج اللغوي" — suggestion to enable upgrading the language model)
• Abdellatif replied: yes (asked what monthly price would be fair; Majed said he lacks the knowledge and suggested asking AI to compute it from database/model/stack costs)

Discovered Qaf through a WhatsApp video and is delighted. Suggests a paid tier on a stronger model:

ولكن بما أن التطبيق مجاني وبلا إعلانات فربما انكم تستخدمون نموذجا لغويا عبر ال API ليس هو الأحدث، لذلك أقترح إتاحة نموذج لغوي أقوى مثل GPT-5.6 Sol أو غيره باشتراك مدفوع لتحسين دقة النقول والإجابات، مع العلم أني لم ألحظ أي شيء غير دقيق حتى الآن.

But since the app is free and ad-free, you are perhaps using a language model via the API that is not the latest; therefore I suggest offering a stronger language model such as GPT-5.6 Sol or another via a paid subscription, to improve the accuracy of quotations and answers — noting that I have not observed anything inaccurate so far.

محمود بن احمد الهاروني (Mahmoud bin Ahmed Al-Harouni) — mahmoudahmedz967@gmail.com

• Date: 2026-08-13
• Language: Arabic
• Thread: 19ffb07ea436d4f5 (subject: a "comprehensive letter" — thanks, technical problems, then suggestions)
• Abdellatif replied: dataset marks the thread replied, but the captured messages are internal forwards to Rahma Fateen ("he's a blind man and wrote this... you need to take it seriously"; Abdellatif asked her to draft a reply) — no direct reply to the sender appears in this thread.

A blind user (Android + TalkBack/Jieshuo; his accessibility bug reports are tracked separately). Under "suggestions for future development" he proposes tiered plans:

لذلك، إذا قررتم مستقبلًا تقديم خطط مدفوعة، أقترح عدم الاكتفاء بخطة مدفوعة واحدة، وإنما تقسيم الاشتراك إلى عدة مستويات تناسب شرائح المستخدمين المختلفة، مثل: * Premium: أعلى حدود للاستخدام وأوسع إمكانات البحث والتحليل. * خطة متوسطة: إمكانات جيدة وحدود استخدام أقل بسعر أكثر ملاءمة. * Lite: إمكانات أساسية بسعر منخفض. * خطة مجانية: تظل متاحة للجميع، ولكن بحدود استخدام أو بعض القيود على الميزات المتقدمة.

Therefore, if you decide in the future to offer paid plans, I suggest not limiting yourselves to a single paid plan, but splitting the subscription into several tiers suiting different user segments, such as: Premium — the highest usage limits and the broadest search and analysis capabilities; a middle plan — good capabilities and lower usage limits at a more affordable price; Lite — basic capabilities at a low price; a free plan — remaining available to everyone, but with usage limits or some restrictions on advanced features.

He stresses keeping a free plan so students and researchers who cannot afford a high subscription retain access, while giving the project an income stream. He then proposes moving from "Islamic research assistant" to a "complete Islamic assistant":

• أن يستطيع المساعد البحث عن مواقيت الصلاة بحسب موقع المستخدم، مع ما يتصل بها من مسائل وأحكام.
• وإذا أمكن مستقبلًا دمج قاف مع خدمات Quran.com أو الاستفادة من خدماته وواجهاته البرمجية، فسيكون ذلك إضافة ممتازة

That the assistant can look up prayer times based on the user's location, along with the related questions and rulings. And if it becomes possible in the future to integrate Qaf with Quran.com's services or benefit from its services and APIs, that would be an excellent addition.

Details / repro

• Majed's premise is a guess about the backend ("perhaps you are using a model via the API that is not the latest"); he explicitly reports zero observed inaccuracies. The ask is really "let me pay for the best model."
• Pricing signal: when asked directly, Majed had no number in mind — no willingness-to-pay data yet; he suggested modeling costs (database, chosen premium model, stack) to derive the price.
• Al-Harouni's rationale for tiers: AI market economics — model inference, infrastructure, and continuous development costs make an entirely free product unsustainable long-term.
• Al-Harouni is a screen-reader user; any premium/paywall UI must be accessible (see his separate accessibility reports in the same thread).

Threads

• 1a006f06be307f8e — Majed Al-Harbi: paid subscription unlocking a stronger LLM (e.g. GPT-5.6 Sol) for quote/answer accuracy
• 19ffb07ea436d4f5 — Mahmoud Al-Harouni: Premium/mid/Lite + free tier structure; full Islamic assistant (prayer times by location); Quran.com API integration