Start Earning →

For Providers

Share compute.
Earn MYAI.

Run the agent on any machine with a GPU. It polls for jobs 24/7 and earns MYAI tokens for every inference chunk it completes.

Step 1

Install the agent

One command installs the agent, sets up Ollama, and registers your GPU with the network. Works on Linux and macOS (Windows is not supported yet). No sign-up needed.

curl -fsSL https://myaitoken.io/install | bashLinux · macOS

Step 2

Pull models

More models = more jobs. Pull the ones that match your hardware. The agent auto-detects what you have and only accepts jobs it can run.

Recommended — Best earnings/hardware ratio

llama3.24GBFastDefault — runs on anything
ollama pull llama3.2
phi410GBFastBest quality per watt · Mac Mini ideal
ollama pull phi4
deepseek-r1:7b6GBMediumHigh demand · reasoning tasks
ollama pull deepseek-r1:7b
mistral6GBFastReliable, widely requested
ollama pull mistral
qwen2.5-coder:7b6GBFastCode generation — dev market
ollama pull qwen2.5-coder:7b

High Performance — More VRAM, higher earnings

llama3.1:70b40GBSlowNear GPT-4 quality
ollama pull llama3.1:70b
deepseek-r1:70b40GBSlowBest reasoning available
ollama pull deepseek-r1:70b
mixtral:8x7b24GBMediumMoE · great quality/speed tradeoff
ollama pull mixtral:8x7b
qwen2.5:72b48GBSlowTop tier multilingual
ollama pull qwen2.5:72b
gemma2:27b20GBMediumGoogle · premium quality
ollama pull gemma2:27b

Embeddings — RAG pipeline demand

nomic-embed-text1GBFastMost popular for RAG
ollama pull nomic-embed-text
mxbai-embed-large1GBFastHigh-dimensional embeddings
ollama pull mxbai-embed-large
Pull as many as your disk allows. The agent reports its available models on every heartbeat — job routing is automatic.

Step 3

Earn

01
Poll
Agent polls coordinator every 5s for available jobs matching your models
02
Run
Claims a chunk, runs inference locally via Ollama, returns result
03
Earn
MYAI accrues to your pending balance per verified job; weekly on-chain payouts to your Base wallet start once switched on

Provider Economics

What can I earn?

Earnings are calculated per verified inference job from output tokens, the model's compute units, your tier multiplier and your reliability score, then settled daily against per-tier gross-accrual caps (Datacenter 200K · Consumer 100K · Browser 50K · Mobile 25K = 375K/day). The tier caps bound how much gross earnings each hardware class can accrue; what is actually paid is bounded by the weekly on-chain payout cap — 200,000 MYAI/week in Year 1. Payout status: on-chain payouts have not started yet. Every verified job is already recorded to the provider's pending balance; the weekly payout described here is the schedule that will pay those balances once it is switched on.

Tier
Hardware
Typical nodes
Daily cap (default)
Tier multiplier
Datacenter
Multi-GPU server
Linux, cloud instances
Highest cap, best jobs
200,000
0.7×earnings multiplier
Consumer
Laptop / desktop GPU
Mac, Linux
RTX 3060+, Apple Silicon
100,000
1.0×earnings multiplier
Browser
Browser-native agent
WebGPU / WASM, in-tab
Zero install — runs while the tab is open
50,000
1.0×earnings multiplier
Mobile
Smartphone / edge device
iOS, Android
Smaller jobs, lower cap
25,000
1.2×earnings multiplier

Per-job earnings

earnings per job = compute units × (output tokens ÷ 1,000) × tier multiplier × reputation multiplier

Llama 3.2 · 3B
0.5 compute units / 1K tokens
Qwen 2.5 · 7B
1.0 compute units / 1K tokens
Qwen 2.5 Coder · 14B
2.0 compute units / 1K tokens
Qwen 3.5 / 3.6 · 27B
4.0 compute units / 1K tokens
Any other model
1.0 compute units / 1K tokens

Reputation multiplier = 0.5 + reliability score, so 0.5× for a new or unreliable node up to 1.5× for a fully reliable one. There is no per-job minimum: a short completion earns proportionally less.

Tier caps bound daily gross accrual across all providers in a tier — if a tier exceeds its cap, that tier is scaled pro-rata; they are configured defaults and do not halve. Once live, payouts are made weekly (Sundays 02:00 UTC) under a network-wide weekly cap that follows the halvening schedule: 200,000 MYAI/week in Year 1, 100,000 in Year 2, then halving every 2 years (figures are coordinator defaults). If pending earnings exceed the weekly cap, every provider is paid pro-rata and the remainder rolls into the following week. USD value depends on MYAI market price.

Live Network

Provider Leaderboard

Loading...

Powered by GET /v1/agents/leaderboard · cached 1 h

Track earnings on Telegram

Connect @MyAIEarningsbot to get real-time earnings, job status, and provider stats right in Telegram.

Open Bot

API Endpoints

GET/api/v1/agents/gpu/onlineList all online GPU providers
POST/api/v1/jobs/parallelSubmit a parallel inference job
GET/api/v1/jobs/parallel/:idGet job status and result
POST/api/v1/agents/:id/dispatchDispatch directly to a specific agent
GET/api/v1/agents/gpu/lookup?api_keyLook up your agent by API key
GET/v1/agents/leaderboardProvider leaderboard (reputation/earnings/jobs)

Base URL: https://api.myaitoken.io