A shiver is the collective noun for sharks — and a literal description of how our compute runs. Every participating GPU contributes to a cold-running mesh that serves inference in real time.
Got an idle 3090 / 4090 / A100? Install the SHARK client, configure your thermal cap, run it cold, earn $SHIVER. Payouts every epoch based on uptime × throughput.
READ THE DOCS →Send your prompts to the Shiver API — pay-per-token in $SHIVER. Around 30% cheaper than centralized providers, no per-minute rate limits, no content guardrails.
API REFERENCE →GPUs run capped at a user-set temperature ceiling, protecting hardware longevity. The client auto-detects your card and proposes a safe cold target.
Requests route to the lowest-latency cold node that can serve the requested model. No central dispatcher — peer-to-peer with onchain attestation for billing.
Every epoch, contributors receive $SHIVER proportional to their served-token count, weighted by quality score from peer verification.
You leave any time. Drain your $SHIVER, switch off the client. Hardware is yours. No staking penalty, no vesting on rewards already earned.