NanoGPT Subscription

No. Creating or using multiple accounts to bypass limits can result in suspension; contact support if you need help consolidating access.

No. Subscriptions are for one person only and cannot be shared, split, resold, or used as a pooled account for multiple people. Each person needs their own account and subscription.

Yes! Your subscription works seamlessly across both our website and API. Use the same quota whether you're chatting on the web or integrating with your applications.

Usually not as your primary programming setup. Pro includes eligible coding models, but web search is not included in subscription coverage, and coding harnesses or agent loops can send large prompts, repo context, tool results, and retries that use included input tokens quickly. For heavy coding workflows, pay as you go or balance overage is usually the safer default.

Yes, but only on a best-effort basis. /v1/completions is a legacy compatibility endpoint, and some underlying providers only support chat-style APIs, so behavior can be limited or inconsistent depending on the model. For the most reliable behavior, use /v1/chat/completions.

No, it's not. You can see on our API page or Pricing page what the context size for the models is; we generally offer the full context size for every model. The context size is the same across pay as you go and subscription, so to be clear no, we do not limit the context size of the models if you subscribe.

Conversations are stored on your device by default unless you enable optional sync or sharing features. We apply the strictest retention and deletion settings providers make available and prioritize providers with no-training/no-log commitments for open-source routes. Final retention and logging behavior remains provider-specific.

Yes! The subscription gives you free access to open-source models, but you can still use premium models like GPT-5, Claude, Gemini and others with your regular balance.

No. Subscription-covered models use auto routing. If you choose a specific provider, that request uses pay-as-you-go pricing from your regular balance. On the API, provider selection is blocked when "Enable paid models on API" is turned off.

All other text models that are not already included in the subscription. This includes TEE variants and proprietary models like ChatGPT (OpenAI), Gemini (Google), and Claude (Anthropic).

If you hit a subscription token or image cap (for example the weekly included input-token cap or daily free image cap), included usage pauses until that limit resets unless you enable "Use balance after limits," which lets you continue on your regular balance.

Unless it's on the included open-source model list below or one of the image models mentioned here, likely not. GLM 5.1, Xiaomi MiMo V2.5, Xiaomi MiMo V2.5 Pro, Kimi K2.6, MiniMax M2.7, and DeepSeek V4 Pro are included. GLM 5.1 and DeepSeek V4 Pro consume included input tokens at 2x. Video generation, voice, and extra/custom tools like web search or extended memory are also pay-as-you-go.