Light API Frequently Asked Questions (FAQ)
Clear answers about free and paid modes, network limits, Telegram and website accounts, API keys, desktop installers and data privacy. All sections are server-rendered and work without JavaScript.
General & Model Catalog
What is Light API and how can I use it?
Light API provides access to language models through a browser chat, the Light AI terminal assistant, a Telegram bot, and an OpenAI/Anthropic-compatible HTTP API gateway. You can chat directly on the web or connect your IDE and terminal with a personal API key.
Which models are currently connected?
In the current server configuration, 14 of 20 catalog models are active: GPT 5.6 Sol, GPT 6 Astra, Claude Sonnet 5.5, Opus 5.5 - supplier 1, GLM 5.3, Kimi K3, DeepSeek V4 Pro, DeepSeek V4.1 Flash, Qwen 3.8 Flash, MiMo V2.6 Pro, Grok 4.7, Gemini 3.8 Flash, HY4 Preview, Gemini 4 Argon. See /en/models for all API IDs, statuses, and rates.
Do I need an account to try the chat?
No. Guests can send up to 10 free messages per day without signing in. Signing in with Telegram or a website password raises the daily free chat allowance to 50 messages.
Free Tier & Rate Limits
What are the free chat limits?
Up to 10 messages per day for guests and up to 50 messages per day for signed-in users. Prompts are capped at 2024 tokens, outputs at 512 tokens, and context retains the last 5 messages.
Why do free limits also depend on IP and network?
The shared free budget is protected by separate account, IP and subnet limits. A shared network limit can apply before your personal allowance is reached. The administrator can adjust these limits; short burst windows also apply. Daily counters reset at 00:00 UTC.
Do free limits apply to paid chat or API keys?
No. Paid chat and API keys use the selected model limits and your personal wallet balance (or key spending cap), and paid chat preserves full conversation history.
Pricing, Wallets & Referrals
Are the model prices on the site confirmed?
Current catalog rates are marked as verified in the configuration. Every model card explicitly states whether its rate is a draft or verified.
What is the difference between Telegram sign-in and password accounts?
Signing in with Telegram connects to the shared wallet used by the Telegram bot so your balance and usage appear in both places. Bot and website conversations are stored separately. Registering with a login and password on the website creates a separate account with an independent wallet.
How does the referral commission work?
You receive 10% commission from confirmed genuine top-ups made by referred users. Test credits never generate commission, refunds adjust commission balances, and withdrawals are handled manually through support.
API Keys, Installers & Privacy
How do I connect VS Code, Cursor, Claude Code, or OpenCode?
Once your wallet is activated, create a key on the connection page with a USD spending cap, expiration, and allowed models. Download the ready ZIP installer for Windows (Install.cmd) or macOS (Install.command), or configure the endpoint manually. For Cursor, the installer configures the Continue extension; native built-in Cursor chat requires manual setup and public HTTPS.
What happens if a stream disconnects or my balance is low?
Before dispatching a paid text request, the gateway reserves the upper-bound token cost. If your wallet balance or key cap is too low, the request is rejected before calling the provider at zero cost. Interrupted streams settle only confirmed usage, and the crash-recovery journal refunds unsettled reservations if the gateway restarts.
How are conversations and API keys stored?
Website chats are stored until you delete them manually. Paid chat retains conversation history; free chat retains up to the last 5 messages. Deleting a chat removes its messages and title from the live store. The Light API is strictly no-log for content: prompt and response text is not saved. Only operational and billing metadata remains: time, model, tokens, cost, key, IP address and processing status. Deleting a conversation from the live store does not immediately remove its previous data from existing backups. No-log applies to storage by Light. Requests are sent to the selected provider to generate an answer; providers and Telegram have their own data handling policies.