FAQ

Which models are available?

Use the model list returned for your TokenHub key. Public snippets should use a model that appears in that list.

Is token-hub really OpenAI-compatible?

Yes for Chat Completions-compatible clients. Point your OpenAI SDK at https://llm.sandboxclaw.com/v1, use Authorization: Bearer $TOKENHUB_KEY, and request moonshot-v1-8k.

Can Claude Code connect?

Yes. Configure Claude Code-style clients with:

export ANTHROPIC_BASE_URL=https://llm.sandboxclaw.com
export ANTHROPIC_API_KEY=$TOKENHUB_KEY

The client reaches /v1/messages.

Does Codex work today?

Codex should use TokenHub as a Responses provider. End-to-end behavior depends on the model enabled for the API key.

Can DeepSeek-TUI connect?

Yes. DeepSeek-TUI reads OpenAI-compatible endpoint settings from environment variables. Use:

export DEEPSEEK_API_KEY=$TOKENHUB_KEY
export DEEPSEEK_BASE_URL=https://llm.sandboxclaw.com/v1
export DEEPSEEK_MODEL=moonshot-v1-8k

Where am I billed, and in what currency?

Billing runs through Stripe. Public checkout is currently USD-only, with topups starting at $5. Credits post to your account shortly after checkout and stay valid for 12 months. There is no subscription and no monthly minimum.

Do I need a fixed API key prefix?

No. Use the full TokenHub key generated by the console as a bearer token. Do not depend on a fixed prefix. Keep the key server-side and rotate it if you think it leaked.

Are my prompts used to train models?

No. token-hub does not use prompt or completion content for training, analytics resale, or model improvement. Requests are forwarded to the selected upstream model channel. Operational metadata such as status code, token count, latency, model ID, and request ID may be kept for debugging and abuse detection. Upstream providers have their own data policies while processing requests.

What are the default rate limits?

The public default is 60 requests per minute per key and 600 requests per minute per account. Burst above that returns 429 rate_limit_exceeded with a Retry-After header when available. Higher limits are granted on request - email support@sandboxclaw.com with your use case and expected RPM.

What happens when an upstream channel is down?

We return 502 upstream_error or 504 upstream_timeout with the upstream reason when available. The request is not billed when the gateway rejects it before a completed upstream response. Retry transient errors with exponential backoff. If another enabled model channel is available for your account, your application can choose to retry there.

Can I get a refund?

Yes. Unused credits are refundable within 7 days of the original topup - email support@sandboxclaw.com with the topup ID. After 7 days, credits stay on your account and expire 12 months from purchase.

Can I revoke a key I think was leaked?

Go to /keys, click Revoke next to the affected key. Revocation is instant for new requests. Rotate clients to a new key, then delete the old row.

Is there an SLA?

Not in MVP. The service is best-effort. Enterprise SLAs, dedicated capacity, or custom routing can be discussed after the public routes are stable.