What bring-your-own-key actually costs: the real math

Every AI tool you use is paying for tokens somehow. The $20 subscription bundles hide the meter behind a flat fee and a rate limit; API accounts put the meter in front of you and bill per million tokens. Dracon is built on the second shape: you bring your own provider keys, our tools are the interface, and the token bill lands on accounts you own. This article is the arithmetic of that choice, using only prices from our provider catalog and our own pricing page.

What the subscription shape costs

Our catalog tracks 42 AI providers, and the consumer subscription tiers cluster exactly where you'd expect: ChatGPT Plus at $20/month, Claude Pro at $20/month, Google AI Pro at $19.99/month, Mistral Pro at $14.99/month, SuperGrok at $30/month. These are good products, but notice the unit: you are buying a monthly allowance of messages, not tokens. The catalog's own usage-limit fields show what that means in practice: ChatGPT Plus lists roughly 20 messages per 5-hour window on base Codex models, and the tiers above it are mostly bigger multipliers of the same window. If you use two of these bundles, a common combination, you are at $35 to $50 a month before any API spend.

What the BYOK shape costs

Now the other shape. Take the cheapest frontier-class API price in our catalog right now: DeepSeek V4 Flash, listed from $0.14 per million input tokens and $0.28 per million output tokens as of its 2026-07-31 public beta. A month of heavy personal chat, say 30 long conversations a day, works out to roughly 1 million input and half a million output tokens. That is $0.28. Add Dracon Omni at the $10 founding price and your month costs $10.28. Even at ten times that volume you are at $12.80, still under a single $20 bundle. And if you route through the free tiers instead (Groq serves up to 14.4K requests a day with no card, OpenRouter gives 50 free-model requests a day, Google AI Studio has a free daily quota), the token line is literally $0 and the month costs the $10 sub alone.

Why our side of the bill can be $10

The reason is not generosity, it is cost structure. When you chat on /ai-hub/chat, your browser calls the provider directly with your key; we never proxy the traffic, so we never pay for your tokens. Our internal operating plan states the resulting number plainly: marginal cost per subscriber is approximately $0. Static pages, object-storage streaming, and browser-direct AI mean one more subscriber costs us nothing measurable, so the sub can be priced like content, not like compute. The $10 founding tier (locked for as long as you stay subscribed; $15 standard later) buys everything on dracon.uk, everything we ship later, and zero markup on any provider key you plug in.

The honest catches

• You manage your own keys. Signing up for a provider, generating a key, and pasting it into a tool is five minutes once, but it is five minutes a bundle does not ask for.
• Cheap tokens are not free tokens at the frontier. If you insist on the most expensive flagship models for everything, API billing can pass $20/month; the math above works because you pick the model per task, which is the whole point of BYOK tooling.
• Rate limits move both ways. Bundles cap your messages per window; free API tiers cap requests per day. Our free-tier roundup tracks which free tiers are actually usable, because that answer changes.
• The $20 bundle is still the right buy if you want one bill, one app, and zero setup. We are the shape for people who would rather own the meter.