Pricing

Pay-as-you-go credits. $1 in free credits on signup. No subscription, no expiry, no minimum. Self-serve top-ups go up to $100 per Stripe session — for larger invoices, email support@fredcode.net.

Full retail, all the time. We don't pass through DeepSeek's promotional or off-peak discounts. Your bill is the same regardless of when you call — predictable, auditable, and immune to mid-month rate changes.

Flash

deepseek-v4-flash · Default. Fast, cheap, good enough for almost anything.

Token typePer 1M tokens
Input (uncached)$0.35
Input (cached)$0.0070
Output$0.70

Pro

deepseek-v4-pro · Heavier reasoning. ~12× the output cost — switch with /pro in the CLI.

Token typePer 1M tokens
Input (uncached)$4.35
Input (cached)$0.04
Output$8.70

Fred charges a 2.5× margin over DeepSeek's published rates. Even with the margin, Fred remains ~10× cheaper per session than Anthropic Sonnet or OpenAI direct on equivalent token loads — see the estimator below.

Estimate your monthly cost

Drag the sliders. Numbers come from the same lib/pricing.ts the proxy bills against — no fudge factor.

Input tokens / turn50K
Output tokens / turn10K
Turns per day20
Cached input40%
Fred / month
$10.58
Claude Sonnet 4 / month
$147.60
GPT-4o / month
$120.00

Computed as turn-cost × turns/day × 30 days, using each provider's published per-token rate. Cache discount applied to input only.

Web research

Tavily-backed web_search and web_fetch tools. Off by default — flip on at /settings. Bills against the same credits balance.

ActionYou pay
web_search · basic$0.020
web_search · advanced$0.040
web_fetch · basic · per 5 URLs$0.020
web_fetch · advanced · per 5 URLs$0.040

URL counts ceil to the next 5: fetching 6 URLs bills as 2 batches. Same 2.5× margin as token usage. See the docs for the toggle, permission flow, and limits.