Introduction
InferaNet is a decentralized GPU compute marketplace built on Solana. It connects AI developers who need GPU power with hardware providers who have idle capacity, creating a transparent and efficient market for AI compute.
Who Uses InferaNet
Developers and AI teams rent GPU compute on-demand for inference, training, rendering, and agent workloads without cloud vendor lock-in or long-term contracts.
GPU providers contribute idle hardware to the network, set their own pricing, and earn USDC for completed and verified workloads.
How It Works
The marketplace operates as a two-sided compute economy. Renters browse available GPUs filtered by class, VRAM, region, and price. Providers register hardware, run benchmarks, and can stake IFN for reputation. Rental settlement happens in USDC on Solana with per-second billing granularity.
Every workload is verified through the Proof-of-Compute layer, which checks execution integrity, uptime, and resource usage before releasing payment to providers.
Quickstart
Get up and running with InferaNet in five steps.
1. Connect Your Solana Wallet
Click Connect Wallet in the top navigation or dashboard. InferaNet supports Phantom, Solflare, and Backpack wallets.
2. Fund Wallet with USDC
Fund your connected Solana wallet with native USDC-SPL. GPU rental payments are sent in USDC on Solana.
3. Choose a GPU
Navigate to the GPU Marketplace and browse available hardware. Filter by GPU class, VRAM, region, provider score, and price.
4. Configure Your Workload
Go to Deploy Workload and specify your workload type, container image, runtime limit, and environment variables.
5. Deploy and Monitor
Click Deploy to submit your job. Monitor progress in Active Jobs where you can see runtime, cost, proof status, and logs.
Rent a GPU
The GPU Marketplace lists all available hardware from registered providers across the network.
GPU Selection
Each listing shows the GPU model, VRAM capacity, supported workload types, region, hourly pricing, availability status, and provider reputation score.
Pricing
Prices are displayed per hour for readability, but actual billing is calculated per second on Solana. This means you pay only for the exact compute time consumed.
Regions
GPUs are distributed across global regions including Singapore, Tokyo, Frankfurt, Los Angeles, New York, London, Amsterdam, Seoul, and Jakarta. Select the region closest to your users for lowest latency.
Provider Score
Provider score ranges from 0-100 and reflects uptime history, job completion rate, verification success, and stake amount. Higher scores indicate more reliable providers.
Deploy Workloads
InferaNet supports five workload types, each optimized for different compute requirements.
AI Inference
Run pre-trained models for real-time inference. Supports LLMs, vision models, embedding models, and classification pipelines. Recommended GPUs: RTX 4090, A6000, H100.
Model Training
Fine-tune or train models with distributed GPU support. Checkpoint saving and resumable training sessions. Recommended GPUs: A100, H100.
Rendering
GPU-accelerated rendering for images, video, and 3D assets. Supports Blender, Arnold, and custom rendering pipelines. Recommended GPUs: RTX 4090, A4000.
Agent Runtime
Long-running autonomous agent sessions with persistent compute and state management. Recommended GPUs: RTX 3090+.
Custom Container
Deploy any Docker-compatible container with full root access and custom environment configuration.
Model Endpoints
InferaNet provides pre-configured model deployments for popular AI models. Each endpoint is optimized for the recommended GPU class.
Available Models
Llama 3 (70B) for text generation and reasoning. Stable Diffusion XL for image generation. Flux for photorealistic synthesis. Whisper for transcription. Mistral for efficient inference. Custom endpoints for any containerized model.
Provider Setup
Hardware Requirements
Minimum: NVIDIA GPU with 8GB+ VRAM, 16GB RAM, 100Mbps internet, Ubuntu 22.04+. Recommended: 24GB+ VRAM, 64GB RAM, 1Gbps internet.
Node Client
Install the InferaNet node client to connect your hardware to the network:
curl -fsSL https://get.inferanet.io | bash
inferanet provider init
inferanet provider register
Benchmarking
The benchmark suite tests GPU compute performance, memory bandwidth, and network throughput to establish a verified performance profile.
Pricing
Set your hourly rate based on GPU class and market conditions. The platform provides suggested pricing based on current network averages.
Staking
Stake IFN to build reputation and access premium job tiers. Rental payments remain priced and paid in USDC.
Proof-of-Compute
The Proof-of-Compute (PoC) layer verifies that providers correctly execute workloads before releasing USDC payment.
Job Attestation
Each job produces a cryptographic attestation containing resource usage metrics, execution logs, and output hashes.
Verification
The verification layer cross-checks attestations against expected resource consumption and output validity.
Reputation
Providers accumulate reputation based on verification success rates, uptime, and stake amount. Reputation directly affects job allocation priority.
Slashing
Providers who submit false attestations or fail verification can face automatic slashing of staked IFN collateral.
Billing
Per-Second Billing
All compute usage is billed per second. Prices displayed per hour are for comparison only. Actual charges are calculated as: (seconds used × price per hour) / 3600.
Escrow
Before deploying a workload, funds are placed in a Solana escrow smart contract. Unused funds are returned when the job completes or is cancelled.
Usage Receipts
Every completed job generates a detailed receipt with GPU class, runtime, cost, proof hash, and Solana transaction signature.
API Reference
The InferaNet API provides programmatic access to the GPU marketplace, job management, and billing.
Authentication
All API requests require an API key in the Authorization header:
Authorization: Bearer ifn_sk_your_api_key
Endpoints
GET /v1/gpus
Returns a list of available GPUs with pricing and provider information.
curl -H "Authorization: Bearer ifn_sk_..." \
https://api.inferanet.io/v1/gpus
POST /v1/jobs
Create a new compute job.
curl -X POST https://api.inferanet.io/v1/jobs \
-H "Authorization: Bearer ifn_sk_..." \
-d '{"gpu":"rtx4090","type":"inference","image":"user/model:latest"}'
GET /v1/jobs/:id
Retrieve the status and details of a specific job.
curl https://api.inferanet.io/v1/jobs/job_8341 \
-H "Authorization: Bearer ifn_sk_..."
POST /v1/providers/register
Register new provider hardware on the network.
GET /v1/billing/usage
Fetch billing history and usage metrics for the authenticated account.
CLI Reference
The InferaNet CLI provides command-line access to all marketplace and provider functions.
Installation
curl -fsSL https://get.inferanet.io/cli | bash
Commands
inferanet login # Authenticate with API key
inferanet gpus list # List available GPUs
inferanet gpus list --region tokyo # Filter by region
inferanet deploy \
--gpu rtx4090 \
--region singapore \
--image user/model:latest # Deploy a workload
inferanet jobs list # List your jobs
inferanet jobs status job_9281 # Check job status
inferanet jobs stop job_9281 # Stop a running job
inferanet provider register # Register as provider
inferanet provider benchmark # Run benchmark suite
inferanet billing usage # View billing history
inferanet billing receipts # Download receipts
Token Utility
IFN can remain the native utility token for reputation, governance, and priority access, while GPU rental settlement uses USDC.
Compute Payments
All GPU workloads are priced and settled in USDC. Escrow contracts hold funds during execution and release payment upon verification.
Provider Rewards
Providers earn USDC for completed and verified compute jobs. Reward amounts are based on GPU class, runtime, and provider tier.
Staking
Providers stake IFN to build reputation and access premium workload tiers. Staked tokens serve as collateral against fraudulent behavior.
Governance
Token holders can participate in protocol governance votes covering fee structures, network parameters, and upgrade proposals.
Fee Burns
A portion of marketplace transaction fees are permanently burned, creating deflationary pressure as network usage grows.
FAQ
What is InferaNet?
InferaNet is a decentralized GPU compute marketplace built on Solana. It connects developers who need GPU power for AI workloads with providers who have idle hardware capacity.
Why Solana?
Solana provides sub-second finality, low transaction costs, and high throughput — essential for per-second billing settlement and real-time compute payments.
How are providers paid?
Providers earn USDC for each completed and verified workload. Payment is released from escrow once the Proof-of-Compute layer confirms execution integrity.
How does Proof-of-Compute work?
Each workload generates a cryptographic attestation with resource usage metrics and output hashes. The verification layer validates these proofs before releasing provider payment.
Can I rent GPUs without long-term contracts?
Yes. InferaNet uses per-second billing with no minimum commitments, no contracts, and no lock-in. Pay only for the compute time you use.
What is USDC used for?
USDC is used for GPU rental payments and provider rewards. IFN can still be used separately for reputation staking, governance, and priority access.
How do I become a provider?
Install the InferaNet node client, register your GPU hardware, run the benchmark suite, set your USDC pricing, stake IFN if required, and go live to accept workloads.
InferaNet