🧭 Page Guide
New here? This section explains what you see on this page and how to use it. You can turn this off anytime from My Profile.
🤖 AI Settings
This page controls everything about AI on the platform — your preferences, your models, your spending limits, and full transparency on how much each AI action costs. Take your time here — understanding this page helps you get the most out of your AI credits.
Title: AI Settings with subtitle "Manage your AI preferences and view usage." ➡️ On the right, a green badge ✅ AI Enabled confirms that AI features are active on your account.
Four stat cards showing your current AI consumption:
📤 This Month Requests — how many AI requests
you've made this calendar month.
🪙 This Month Tokens — total tokens consumed
this month (input + output combined).
📈 Total Requests — lifetime count of all AI requests
since account creation.
💰 This Month Cost — the total credits consumed
by AI this month. This comes from your AI Quota pool.
Two cards showing your account's AI limits:
📋 Requests per Day — maximum number of AI requests
you can make per day (e.g. 500). This prevents accidental
runaway usage.
📋 Monthly Token Limit — maximum tokens you can consume
per month. Shows "Unlimited" if no cap is set.
Below the limits, Features Available shows
which AI capabilities are enabled:
✅ Streaming (real-time AI responses),
❌ Vision (image analysis — not available),
✅ Audio (speech-to-text via Whisper).
The AI model used for text conversations (the AI assistant,
prompt generators, AI wizards). Dropdown with options:
🟢 Claude Haiku 4-5
[Recommended] —
fast, affordable, great for everyday tasks. Best balance
of quality and cost.
🔵 Claude Sonnet 4-6 — more capable,
better at complex reasoning. Costs ~3x more than Haiku.
🟣 Claude Opus 4-6 — the most powerful model.
Best for complex analysis and creative work.
Costs ~5x more than Haiku.
🟠 GPT-4o — OpenAI's model. Alternative
for users who prefer it. Mid-range pricing.
⚪ GPT-4o Mini — OpenAI's lightweight model.
Cheapest text option available.
💡 Start with Claude Haiku 4-5.
It's the recommended default — fast, cheap, and good enough
for most tasks. Switch to a more powerful model only when
you need deeper analysis or more creative output.
The model used when you speak to the AI assistant instead of typing. Currently only one option: Whisper [Recommended]. Whisper converts your voice to text, then the text model processes it. Charged per minute of audio at 0.0090 credits/min.
The model used when the AI reads its response aloud to you.
Two options:
🟢 TTS Standard
[Recommended] —
good quality, half the cost of HD.
🔵 TTS HD — higher quality voice output,
but 2x the cost.
💡 Use TTS Standard unless you specifically need
higher audio quality.
Choose the voice character for text-to-speech output. Multiple options available (e.g. Nova — Warm, friendly female voice). Pick whichever voice you find most comfortable to listen to. This doesn't affect cost — only the sound.
A safety cap on how many credits AI can consume per month.
Set a number in credits — when the limit is reached,
AI stops working until next month.
Set 0 for no limit.
💡 This is your personal safety net.
If you're new, set a limit (e.g. 5 or 10 credits)
until you understand how much AI costs for your usage pattern.
You can always increase it later.
The language the AI responds in by default (e.g. Greek). ⚠️ Important cost note: non-Latin languages (Greek, Chinese, Arabic, etc.) use 2–3x more tokens than English for the same content. This means AI responses in Greek cost more than the same response in English. See the token cost section below for details.
🔵 Auto-apply AI suggestions — when ON,
AI suggestions are applied automatically without asking
for confirmation. When OFF, you review and apply manually.
🔵 Show token usage — when ON,
you see the token count and cost after each AI interaction.
Recommended to keep this ON while you learn the system.
A text box where you can write custom instructions that are sent to the AI with every conversation. For example: "Always respond in simple Greek" or "Focus on 3D printing topics." This is optional — leave it empty if you don't need it.
A token is a piece of text that AI models process. On average, 1 token ≈ 0.75 words (or ~4 characters in English). For example: "Hello world" = 2 tokens, "Artificial Intelligence" = 2–3 tokens.
Every AI interaction has two parts:
🔵 Input = what you send (your question/prompt).
Costs less per token.
🔴 Output = what AI generates (the response).
Costs more per token.
The total cost is: (input tokens × input price) + (output tokens × output price).
Using Claude Haiku 4-5:
Your prompt (~10 words) ≈ ~15 tokens input
AI response (~100 words) ≈ ~135 tokens output
💰 Total cost: ~0.0014 credits
(Input: 0.0000 + Output: 0.0014)
That's extremely cheap. You could send hundreds of messages
like this for a single credit.
1 credit = 10,000 credit units. Prices shown in the pricing table are final user prices (they already include the platform's markup). Your AI Quota balance is measured in credits — the same credits you transfer from MAIN Quota.
Different languages use different amounts of tokens for the same content. Non-Latin scripts (Greek, Chinese, Arabic, etc.) typically use more tokens than English.
~1 token per 4 characters
~0.75 words per token
Baseline cost
~1 token per 1–2 characters
~0.25–0.35 words per token
~2–3x more tokens
⚠️ This means using AI in Greek costs roughly 2–3 times more than the same interaction in English. This is not a platform markup — it's how AI tokenization works. All AI providers (Anthropic, OpenAI) tokenize non-Latin scripts the same way. It's a technical characteristic of how language models process text.
The pricing table at the bottom of the page shows exactly how much each AI model costs. All prices are in credits and include the platform's markup. Here's how to read it:
AI costs are charged from your AI Quota pool (the middle card on the MyQuota page). To have AI credits, you first buy a Usage Pack (credits go to MAIN Quota), then transfer some from MAIN to AI Quota using the "Μεταφορά από MAIN" button. The AI consumes from the AI pool automatically — each action deducts the exact cost shown in the pricing table. If your AI Quota reaches 0, AI features stop until you transfer more from MAIN.