Terms of use

Last updated: 8 October 2026. These terms apply to RedWatts at redwatts.com. Contact support@redwatts.com for service or billing questions.

Your subscription

Core costs $23 USD per month for one fixed daily eight-hour block. Any applicable tax and the total charge are shown before payment. The available blocks are 00:00–08:00, 08:00–16:00 and 16:00–24:00 UTC. Choose one at checkout. Your purchased block repeats daily; local clock times can change with daylight saving. This launch offer does not include annual billing, multiple blocks or mid-period block switching.

Card subscriptions renew automatically each month until cancelled. Cancel through Manage subscription in your dashboard before the next renewal. Cancellation stops future renewals and normally keeps access until the paid period ends. Unused hours do not roll over or create cash credit. Additional API keys share the same hours and one active request channel.

What unlimited means

There is no per-token charge or customer token bundle during your purchased hours. This is shared AI inference access, subject to available processing capacity, variable speeds, model context and output limits, and recovery or safety controls. Requests may be delayed or paused until capacity replenishes. We do not guarantee a minimum token speed, uninterrupted generation throughout the window, or unlimited simultaneous automation.

Inference capacity is funded and shared company-wide, not reserved as a separate monthly allowance for each subscription. Heavy use and concurrent demand can reduce speed or pause new requests while shared capacity recovers; waiting does not guarantee immediate access. A separately configured account safety limit remains a hard limit, and billing, accounting or security reviews can pause access until resolved. These pauses do not add per-token overage charges.

Core includes DeepSeek V4.1 Flash only, using the explicit model ID deepseek/deepseek-v4.1-flash. The selected model is not silently replaced. Access supports text and function tools through an OpenAI-compatible Chat Completions API, with one active request across all keys. Current serving capacity is 262,144 total tokens including input, framing and reserved output space, with up to 65,536 output tokens shared by reasoning, answer text and tool arguments. Default High and explicit Low/High reasoning are supported; omitted output uses 65,536 tokens. The model’s physical 1,048,576 context / 393,216 output metadata is not currently available serving capacity. Explicit output, reasoning and tool intent is preserved: requests that available capacity cannot serve are rejected, not silently clamped or downgraded. A deliberately selected context-fit policy treats output as a ceiling within the physical model context after input and framing; it does not trim input, disable reasoning or expand available serving capacity. Requests are bounded at 8 MiB of UTF-8 JSON. Reasoning effort and workload affect answer timing; partial responses and tool errors require review before continuing. The dashboard provides current compatible setup and permanent-error versus temporary-capacity handling guidance. Existing client configurations are not automatically rewritten. Model availability can change; we will communicate material changes to a paid plan. You receive API service access, not dedicated or reserved GPU hardware.

Payments, cancellation and refunds

Polar acts as merchant of record for card purchases, processes payments and handles applicable transaction taxes. Its checkout and buyer terms also apply. Payment methods, invoices and cancellation are available through Manage subscription.

We do not offer a discretionary money-back guarantee or refunds for unused hours, partial months or changing your mind, except where required by applicable law or Polar's applicable policies. Cancellation alone does not refund a previous payment. RedWatts does not issue automatic refunds. Polar may make its own refund or chargeback decisions, including ending the related subscription.

For duplicate charges, billing errors, failure to supply purchased access, or a refund right under applicable law, email support@redwatts.com with your account email and order reference. Do not send card details or API keys. Nothing here excludes mandatory consumer rights or Polar's authority as merchant of record. A refund and subscription cancellation are separate actions; your dashboard shows whether access remains active.

If we discontinue the service or cannot provide purchased access through the end of a paid period, the affected unused period is eligible for a prorated refund. Card refunds are handled through Polar; wallet refunds follow the ownership verification process below. This exception is separate from a voluntary money-back guarantee.

Agent-native wallet purchases

An agent-native Core purchase authorizes 23 USDC on Base for the exact prepaid calendar-month dates and daily eight-hour block shown in its quote. The block follows your selected IANA timezone. It does not create an automatic renewal, a token allowance or permission for future debits. Another month requires a separate deliberate purchase; changing a quoted window requires a new unpaid quote. All keys for one wallet identity share one active request channel. Wallet and email accounts are separate.

Wallet purchases are sold directly by RedWatts, not through the card checkout merchant of record. USDC is a digital asset, not a promise of a fixed realized USD exchange value. A blockchain transfer cannot be reversed on-chain by cancelling access. For a duplicate payment, billing error, failure to supply access or a refund right under applicable law, contact support with the wallet address, RedWatts order ID and transaction reference. Never send private keys or API keys. Any required wallet refund is handled separately after ownership verification; there is no automatic refund or discretionary money-back guarantee.

Acceptable use

You must be at least 18 and able to enter a subscription agreement. Keep your keys private. Do not publicly share or resell access, bypass request or capacity controls, interfere with the service, or use it unlawfully. Only submit content you are authorized to process. We may restrict access for abuse, security incidents or unpaid charges. AI output can be incorrect; review code and other output before relying on it.

Do not use the service for child exploitation, non-consensual intimate content, deceptive impersonation, privacy or intellectual-property violations, phishing, malware, unsolicited bulk messages, or unauthorized access to systems or data.

Availability

Flagship and Frontier are previews and are not available for purchase. Existing server-managed owner access is separate from a paid subscription and may have different limits and an expiry date.