Skip to content
ObsiKO
English
Get the app

AI Billing Guide (Subscriptions vs Credits)

5 min readUpdated 2026-10-24
On this page

"I'm on ChatGPT Pro — why doesn't this work?" ObsiKO's AI features draw from two different wallets. This page ends the confusion once and for all!

← Back: 16 AI Agent · Home: ObsiKO 101


What's the ⓒ badge?

Features marked with a small ⓒ (credit) badge in menus and dialogs are paid features that call the API keys you entered in Settings. Hover the badge (long-press on mobile) for a quick explanation.

Two wallets — subscription login vs API credits

ObsiKO's AI billing splits two ways. They are completely separate wallets — no matter how premium one plan is, the other holds none of it.

🪪 Subscription login 💳 API credits (ⓒ)
What Account login for Claude Code / Antigravity CLI (agy, Gemini) / Codex CLI API keys in Settings (OpenAI · Gemini · Anthropic)
Billed via Claude: a Pro or Max subscription; Codex: a ChatGPT account (free plan included — limits differ by plan); Gemini (agy): a Google account (free with a weekly limit, more room with Google AI Pro/Ultra) Prepaid credits / usage billing at platform.openai.com etc.
Powers Agent conversation, writing, note organizing (the full agent's brain) Image generation, AI read-aloud (TTS), audio/video transcription, AI spellcheck, AI tidy/summary, Bases AI fill, chat-only mode

Remember one line: "Conversation runs on login; multimedia (images, speech, transcription) runs on credits." And small jobs can run 🆓 free on on-device AI (see the table below).

🔄 Plan names, prices and free limits change often. Check each company's official page for the latest — Claude plans ↗ · Claude Code login guide ↗ · Codex plans ↗ · Antigravity plans ↗ · Gemini API pricing ↗ · OpenAI API pricing ↗

Three common misconceptions

1. "I'm on ChatGPT Pro but transcription fails with 429."
A ChatGPT subscription and OpenAI API credits are separate balances. Transcription and image generation call the API directly, so you need credits at platform.openai.com billing. (Note: if OpenAI transcription fails, ObsiKO automatically retries with Gemini when a Gemini key exists.)

2. "If I ask the agent to draw, does it use my login?"
No! The agent's conversation runs on your login (subscription), but when it draws, the generate_image tool calls the OpenAI/Gemini image API with your API key (credits). Same for read-aloud and cloud transcription. Whether you draw via right-click or by asking the agent, images always use credits.

3. "No credits — can't I use AI at all?"
You can. The desktop full agent works on account login alone — conversation, writing, research, mind map generation (Codex with a free ChatGPT account and Gemini with a free Google account work too, within their limits). On a device with macOS 26 / iOS 26 or later and Apple Intelligence turned on, 🆓 on-device AI handles simple chats, AI spell check, AI polishing and writing-check questions with no key or login (not supported on Android). And the Gemini API has a free tier: grab a key at Google AI Studio and conversation, transcription and read-aloud (TTS) start free within the limits. Image generation isn't on the free tier, though, and content sent on the free tier may be used by Google to improve its products and read by human reviewers (Gemini API terms ↗) → 16-10 AI Sources & Privacy

Billing source by feature

Feature Billed via
AI agent conversation/writing/organizing (desktop) 🪪 Subscription login (or API key)
Chat-only mode (mobile/web) → 16-8 Chat-only Mode 💳 API key
On-device AI — simple chat, AI spell check, AI polishing, writing-check questions, transcript summaries (macOS 26 / iOS 26 or later + Apple Intelligence) 🆓 Free — no key or login (not supported on Android)
AI image generation (right-click AND agent "draw me…") 💳 OpenAI or Gemini key
'Read Note Aloud' (device voice, instant) 🆓 Free — no key needed
AI read-aloud (TTS) audio file creation 💳 OpenAI or Gemini key (Gemini also works on the free tier, within limits)
Voice memo & audio transcription — iPhone/iPad 🆓 Free on the device — no key (with a key you can also choose the cloud)
Video transcription, and audio transcription on other devices 💳 OpenAI (audio) / Gemini (audio·video) key
Post-transcription AI summary · document AI tidy 💳 API key
AI spellcheck 💳 API key (or desktop agent login, 🆓 on-device AI)
Bases ✨ AI fill 💳 API key
Image search · dictionaries · EN spellcheck (ObsiKO server) · JA/ZH spellcheck (LanguageTool) Free (no badge)

💡 Bonus: helper features like transcript summaries, auto memory capture and Bases AI-fill automatically use lighter, cheaper models (haiku / mini / flash tier) — no setting needed, it just saves money. Quality-critical spots like writing-check questions and the main conversation stay on the full model. AI spell check uses a model one step above that (sonnet / default / pro tier) so it finds more errors — it costs a little more than the light models.

💸 Built-in savings: in-app chat and the agent get cheaper the longer you continue the same conversation — Claude remembers the front part and re-reads it at a discount (usually a tenth of the normal price; in the in-app chat the result card shows it as 'N% of input re-read at a discount'). If you pause for more than about 5 minutes before continuing, the memory is cleared and it's read at the full price again, and the first time it's remembered, that part costs 1.25×. And when the desktop Claude agent runs on an API key, a single run stops and tells you once it has spent about $20 — [Continue] carries on but costs more (subscription login isn't billed in dollars, so it isn't affected).

Creating API keys

🔒 Keys are stored on your device only (encrypted on desktop, Keychain on iOS, app settings on Android and web). They are never sent to an ObsiKO server — ObsiKO doesn't even have one!

Troubleshooting

  • 429 "quota exceeded" → that API account is out of credits. Top up or use another provider's key.
  • 401 "invalid key" → typo or revoked key. Issue a new one.
  • Transcription auto-falls back to Gemini when OpenAI fails (if a Gemini key exists).

← Previous: Chat-only Mode (Mobile & Web) · Home · Next: AI Sources & Privacy →