Questions,
answered.

An AI API gateway operated by VD Enterprises, a GST-registered business in India. You call AI models through OpenAI- and Anthropic-compatible APIs and pay from prepaid credits in Indian rupees.

Usually two lines. Point the official OpenAI SDK at https://api.topbilling.in/v1, or the Anthropic SDK at https://api.topbilling.in, and use a Top Billing API key and model ID. A parameter a model cannot honour is rejected with a clear 400, never silently ignored.

The official OpenAI SDKs pointed at https://api.topbilling.in/v1, the official Anthropic SDKs pointed at https://api.topbilling.in (no /v1 suffix; the SDK appends it), plain HTTP such as cURL, and Claude Code. See the SDK guides.

Every model on the Models page. Each lists only what its routes have been verified to do: capabilities, context window, the APIs it answers on and whether it is available now. Model IDs are Top Billing’s own (tb-…).

Yes. Set ANTHROPIC_BASE_URL to https://api.topbilling.in, ANTHROPIC_AUTH_TOKEN to your Top Billing key and ANTHROPIC_MODEL to a model ID from the Models page. The Claude Code guide has the details.

Yes. Run npx @topbilling/cli claude and it writes the Claude Code settings for you. See Set up your tools.

Claude Code’s gateway model discovery only lists IDs containing claude or anthropic, and Top Billing model IDs are neutral (tb-…). Set ANTHROPIC_MODEL to the model ID explicitly.

Yes. Every Claude Code request is billed like any other API request and shows up in Usage with its session id.

No. Use a Top Billing key (tb_…). The gateway rejects credentials that are not Top Billing keys with a 401 and never forwards them.

Add credits on the Billing page with UPI, cards or net banking through Cashfree, our payment gateway. Card, UPI and bank details are entered with the gateway and never reach our servers.

Yes. GST is shown before you pay, and a tax invoice carrying our GSTIN is on the Billing page after every successful payment.

Yes. Every price on this page is in INR before GST. 18% GST is added when you top up, and shown before you pay.

₹100 to ₹50,000 per top-up, by UPI, cards or net banking through Cashfree.

New accounts start with zero credit. A promo code adds credit to a project’s wallet, and it is spent like any other credit.

No. Credits are prepaid in INR and charged per token at the model’s listed price on the pricing page. No subscription, no platform fee, no minimum monthly spend.

Each request reserves an estimate from your credits and settles on the tokens the model reports. Input, output, cache read and cache write tokens are billed separately at the model’s listed price, and whatever the estimate held beyond the final charge is released.

Where a model supports prompt caching, cached input tokens are billed at its cache read and cache write prices. A model with no cache price bills cached input at its input price.

Every response carries an x-request-id header (tb_req_…). The same ID appears on the request’s charge in Usage, so any line of your bill can be traced to one request.

Credits are prepaid, non-transferable and usable only on Top Billing. Refunds follow the refunds and credits policy.

On models with failover enabled, a route that times out, is overloaded or returns a 5xx before producing anything billable hands the request to the next healthy route. The request is reserved once and settled once, so a failover never bills twice.

The request is rejected with 400 capability_not_supported before anything is reserved or charged. Parameters a model cannot honour are never silently ignored.