# StudyFlow Pricing Research

Date: 2026-03-19

This note translates current model pricing into a practical subscription recommendation for StudyFlow.

It assumes the updated product direction:

- public landing page + auth
- document-to-friendly HTML workspace first
- strict grounded chat second
- optional AI study-pack generation third
- optional premium infographic generation for eligible topics

## Core Pricing Insight

The cheapest healthy version of StudyFlow is:

- non-AI transformation by default
- grounded chat with a low-cost model
- optional study-pack generation as the paid upgrade
- premium OCR recovery and infographic generation only as metered premium actions

That is much safer than using a frontier model on every upload.

## Current Reference Pricing

### OpenAI

From OpenAI pricing on 2026-03-19:

- `GPT-5.4`
  - input: `$2.50 / 1M tokens`
  - output: `$15.00 / 1M tokens`
- `GPT-5 mini`
  - input: `$0.250 / 1M tokens`
  - output: `$2.000 / 1M tokens`

### Google Gemini

From Gemini API pricing on 2026-03-19:

- `gemini-2.5-flash-lite`
  - input: `$0.10 / 1M tokens`
  - output: `$0.40 / 1M tokens`
- `gemini-2.5-flash`
  - input: `$0.30 / 1M tokens`
  - output: `$2.50 / 1M tokens`
- `gemini-2.5-pro`
  - input: `$1.25 / 1M tokens` for prompts `<= 200k`
  - output: `$10.00 / 1M tokens` for prompts `<= 200k`

### OpenRouter

OpenRouter currently states:

- pay-as-you-go platform fee: `5.5%`
- no markup on provider token pricing
- fallback attempts are not billed unless the routed model run succeeds

### Payments

Stripe standard pricing reference used for modeling:

- `2.9% + $0.30` per successful transaction

## Why Direct Provider First Matters

If StudyFlow uses OpenRouter for all primary traffic, margins get worse because:

- you pay model cost
- plus OpenRouter platform fee

So the healthier production pattern is:

- direct Gemini or direct OpenAI for primary traffic
- OpenRouter only as fallback

This matches the current app direction already.

## Cost Model Assumptions

These are the operating assumptions used for plan sizing.

### 1. Friendly HTML workspace

This should be non-AI and deterministic.

Expected cost drivers:

- extraction compute
- storage
- indexing

AI cost should be `zero` in the normal workspace-only path.

### 2. Grounded chat turn

Assumed average per chat turn:

- input: `4,000 tokens`
- output: `600 tokens`

Estimated model cost:

- `gemini-2.5-flash-lite`
  - input: `0.004 × $0.10 = $0.0004`
  - output: `0.0006 × $0.40 = $0.00024`
  - total: about `$0.00064` per message

- `GPT-5 mini`
  - input: `0.004 × $0.25 = $0.001`
  - output: `0.0006 × $2.00 = $0.0012`
  - total: about `$0.0022` per message

- `gemini-2.5-pro`
  - input: `0.004 × $1.25 = $0.005`
  - output: `0.0006 × $10.00 = $0.006`
  - total: about `$0.011` per message

### 3. Study-pack generation for a large document

For pricing safety, assume a large document of around:

- `100k input tokens`
- `6k output tokens`

Single-pass baseline cost:

- `gemini-2.5-flash-lite`
  - input: `$0.010`
  - output: `$0.0024`
  - total: about `$0.0124`

- `gemini-2.5-flash`
  - input: `$0.030`
  - output: `$0.015`
  - total: about `$0.045`

- `gemini-2.5-pro`
  - input: `$0.125`
  - output: `$0.060`
  - total: about `$0.185`

- `GPT-5.4`
  - input: `$0.250`
  - output: `$0.090`
  - total: about `$0.340`

But in practice, study-pack generation for large documents will often be chunked and merged.

Conservative real-world budgeting:

- `gemini-2.5-flash-lite`
  - budget about `$0.03 to $0.06` per large study pack
- `gemini-2.5-flash`
  - budget about `$0.07 to $0.15`
- `gemini-2.5-pro`
  - budget about `$0.30 to $0.60`
- `GPT-5.4`
  - budget about `$0.50 to $1.00`

These are safer planning numbers than the direct single-pass math.

## Margin Target

Target margin:

- `30%`

That means total cost should stay at or below:

- `70%` of subscription revenue

### Per-user monthly cost ceiling

- `$5 plan`
  - max total cost: `$3.50`
- `$9 plan`
  - max total cost: `$6.30`
- `$12 plan`
  - max total cost: `$8.40`

### After Stripe fees

Approximate remaining budget after `2.9% + $0.30`:

- `$5`
  - Stripe fee: about `$0.445`
  - remaining for infra + AI + support: about `$3.055`

- `$9`
  - Stripe fee: about `$0.561`
  - remaining: about `$5.739`

- `$12`
  - Stripe fee: about `$0.648`
  - remaining: about `$7.752`

That is still enough room, but only if expensive models are capped.

## Recommended Subscription Design

Do not sell this as unlimited AI.

Use:

- subscription
- included usage caps
- clear monthly allowances
- optional extra credit packs later

The safest user-facing language is:

- documents per month
- grounded chat messages per month
- study-pack credits per month

Avoid exposing tokens to end users.

## Recommended Tiers

### Basic

Recommended price:

- `$5/month`

What to include:

- friendly HTML workspace
- grounded AI chat
- no study-pack generation

Suggested caps:

- `25 document uploads/month`
- `300 grounded chat messages/month`
- `90-day document retention`

Recommended model:

- `gemini-2.5-flash-lite` for grounded chat

Why this works:

- model cost is tiny for grounded chat at this level
- non-AI transform does most of the heavy lifting
- this plan becomes a useful product, not just a teaser

Estimated AI cost at cap:

- `300 × $0.00064 = ~$0.19`

That leaves plenty of room for:

- storage
- extraction
- indexing
- support
- retries

### Plus

Recommended price:

- `$9/month`

What to include:

- everything in Basic
- optional AI study-pack generation
- faster processing queue

Suggested caps:

- `40 document uploads/month`
- `600 grounded chat messages/month`
- `15 study-pack credits/month`
- `90-day or 180-day retention`

Recommended model:

- grounded chat: `gemini-2.5-flash-lite`
- study-pack generation: `gemini-2.5-flash-lite` or `gemini-2.5-flash`

Estimated AI cost at cap:

- chat: `600 × $0.00064 = ~$0.38`
- study packs:
  - `15 × $0.03 to $0.10 = ~$0.45 to $1.50`
- total AI:
  - about `$0.83 to $1.88`

This is very healthy for a `$9` plan.

### Ultra

Recommended price if you want to keep the proposed number:

- `$12/month`

What to include:

- everything in Plus
- deeper study-pack generation
- priority queue
- smarter model for premium actions only
- infographic-style study visual generation for eligible topics

Suggested caps:

- `50 document uploads/month`
- `1,000 grounded chat messages/month`
- `8 deep study-pack credits/month`
- longer retention

Recommended model:

- grounded chat: still `gemini-2.5-flash-lite` or `gemini-2.5-flash`
- deep study-pack generation: `gemini-2.5-pro`

Important constraint:

If Ultra uses the smarter model for every chat turn and every generation, `$12` is probably too tight.

Ultra is healthy at `$12` only if:

- the smart model is reserved for premium actions
- deep study-pack credits are capped

Estimated AI cost at cap:

- chat on Flash-Lite:
  - `1000 × $0.00064 = ~$0.64`
- deep study packs on Pro:
  - `8 × $0.30 to $0.60 = ~$2.40 to $4.80`
- total AI:
  - about `$3.04 to $5.44`

That is workable inside the `$12` budget.

### Infographic generation note

Infographic generation should be treated as a premium credit-based action, not an unlimited side feature.

Reason:

- image generation cost is less predictable than grounded chat
- visual generation can be abused more easily
- premium visuals are a clean upsell for Ultra

Recommended starting policy:

- Ultra only
- `5 infographic credits/month`
- grounded in extracted content only
- reserved for topics where diagrams or concept visuals add real value

## What Not To Do

Do not make:

- `$5` unlimited chat + unlimited documents + unlimited study packs`
- `$12` “smartest model for everything” with no caps

Those sound attractive, but they weaken margin protection and make abuse much more likely.

## Better Pricing Recommendation

The strongest current structure is:

- `Basic — $5`
  - transform + chat
- `Plus — $9`
  - transform + chat + study-pack credits
- `Ultra — $12`
  - transform + chat + deeper study-pack credits with a smarter model on premium actions only
  - separately metered infographic-style study visuals

This works if:

- primary provider is direct Gemini
- OpenRouter is fallback-only
- plan allowances are capped

## If You Want More Generous Ultra

If you want:

- more deep study-pack credits
- smarter chat by default
- higher retention
- larger file allowances

then Ultra should probably be:

- `$15 to $19/month`

That would create a healthier buffer and reduce pricing regret later.

## Recommended Positioning

User-facing plan story should be:

### Basic

- turn documents into a readable workspace
- ask grounded questions

### Plus

- study smarter with AI-generated revision tools

### Ultra

- deeper AI understanding, priority processing, and premium study generation

This is much easier to explain than raw token economics.

## Suggested Perks To Add Later

Safe perks that do not destroy margins:

- longer retention window
- more active documents
- priority processing
- export to markdown or PDF
- saved chat threads
- deeper citation view
- premium workspace themes

Avoid making the main differentiator:

- unlimited expensive model access

## Final Recommendation

If launching soon, I would ship:

- `Basic — $5`
  - 25 uploads
  - 300 grounded chats
  - no study pack

- `Plus — $9`
  - 40 uploads
  - 600 grounded chats
  - 15 study-pack credits

- `Ultra — $12`
  - 50 uploads
  - 1,000 grounded chats
  - 8 deep study-pack credits
  - 5 infographic credits
  - premium model only for deep actions

And I would keep:

- direct Gemini as primary
- OpenRouter as fallback
- non-AI transform as the default path

That is the best balance of:

- product clarity
- cost control
- margin safety
- upgrade logic

## Sources

- OpenAI pricing: https://openai.com/api/pricing/
- Gemini pricing: https://ai.google.dev/gemini-api/docs/pricing
- Gemini quotas: https://ai.google.dev/gemini-api/docs/quota
- OpenRouter pricing: https://openrouter.ai/pricing
- OpenRouter FAQ: https://openrouter.ai/docs/faq
- Stripe pricing: https://stripe.com/pricing