Bittensor mainnet · SN117

Mine with an API key, not a GPU.

Serve media-generation jobs with your own provider account. Everycli sets up a small, attested cloud worker; the provider runs the model, and your worker returns a verified result.

117
mainnet netuid: 59
priced at review: 11
CLI generation providers: 0
GPUs required

#Start with everycli

bash

curl --proto '=https' --tlsv1.2 -LsSf \
  https://github.com/everyframe-studio/everyframe-miner/releases/latest/download/everycli-installer.sh | sh

These guides use everycli v0.8.7. Already installed? Run everycli update. v0.8.3 adds Pika H3 Flex, key validation and API balances. To configure Pika on an existing VM, follow the Pika upgrade steps. Existing automatic controllers continue compatible updates for their current providers. Run everycli ui for the local browser setup and dashboard; terminal commands remain available.

Start with the everycli UI guide to select your hotkey, deploy through the browser and manage keys, balances and model offers.

Already mining? Run everycli health for a color-coded readiness checklist and everycli status for deployment and job-activity tables. Add --json for structured output. Read the health table: all green means current readiness checks passed, not guaranteed traffic or provider funds.

Prefer terminal setup? Use your hotkey that is already registered on SN117:

bash

everycli setup --wallet my-miner --hotkey default --provider fal --max-hourly-usd 0.06
everycli doctor

The interactive wizard checks registration, prompts privately for keys, initializes your profile, then shows a hosting quote for confirmation before creating a paid VM. It does not register on-chain or pay a registration fee. Use a Phala workspace token from Settings → API Tokens, not Private AI → API Keys. No invitation file, miner token or private registry access is required for a new public miner.

The wizard stops if it finds an existing or uncertain deployment and never activates or resumes work automatically. Continue with Deploy your miner for admission, activation and model offers, or use its manual steps instead of the wizard. You need funded hosting and generation-provider accounts. Run everycli help deploy for command-specific options and an example.

everycli UIBrowser setup, balances, batch offers and worker controls. Deploy your minerFrom registered hotkey to your first model offer. Install everycliOne command. Prebuilt binaries for Linux and macOS. RequirementsAccounts, local wallet support and hosting costs. Providers & keysManage keys in the CLI. Check what can serve jobs. RewardsHow accepted work becomes eligible for alpha emissions. How it worksAttestation, provider requests and signed receipts.

#Availability and costs

The 13 October 2026 check confirmed public mainnet onboarding and enabled coordinator routing. This is a dated check, not a guarantee of available work. Check the network dashboard, everycli status and everycli offers for current conditions.

The signed public catalogue reviewed 31 October 2026 contains 100 contracts, 59 priced across 9 providers. Contract counts are not unique-model counts. Current worker permissions, credentials, capacity and explicit offers still determine what can run. See Models & reward targets.

Existing public miners are no longer limited to the old three-model configuration. On your original management device, run:

bash

everycli update
everycli worker-update
everycli offers
everycli doctor

The coordinator supplies a signed update for your existing VM. Expanded permissions take effect after fresh verified attestation; new contracts still need their own explicit offers and provider keys. Do not deploy another VM to upgrade. A paused miner stays paused.

On 13 October, an operator-owned miner using this public CLI flow completed a real website H3 Max Turbo portrait video and Grok image request. This verifies those paths, not every model, other miners, or future availability. See live verification and limitations.

You pay for chain registration, Phala hosting and provider usage. The CLI's hourly ceiling covers the compute rate, not a total budget; storage costs extra. Hosting continues while waiting for work. A submitted generation may cost money even if its result earns no reward. There is no guaranteed traffic, reimbursement or profit.