Request compute. Pay per job.
Ignit3 is a Solana-native GPU bot protocol. Ask for inference, generation, analysis, automation, or market processing straight from a bot — Ignit3 handles pricing, routing, execution tracking, and on-chain settlement. Same program, your dashboard or your Telegram chat.
- <8s
- Job dispatch
- 1.5–3%
- Flat fee
- 420+
- GPUs online
- 100%
- Non-custodial
The network right now
How it works
Submit a task
Tell the bot what you need: inference, generation, analysis. Text or API — same entry point.
Ignit3 routes
The protocol prices the job, matches it to the best GPU, and escrows SOL.
Get results
Output returns to your bot or webhook. Payment settles on Solana automatically.
Every job pays. Pick your side of the network.
Ask a bot · get results
Submit inference, generation, analysis, automation, or market-processing jobs straight from chat. Ignit3 prices the work, routes it to a GPU, tracks execution, and returns your output.
Run a jobRent out idle compute
Connect a GPU and Ignit3 fills it with paid jobs. You earn per completed task in SOL and build a reputation score from uptime, speed, and execution quality.
Become a providerProgrammable execution
AI agents request GPU power, pay per task via x402, receive outputs, and continue their workflows autonomously — no API keys, no accounts, no custody.
Read agent docsOpen infrastructure. No gatekeepers.
An Anchor program on Solana that powers a permissionless GPU marketplace, an on-chain reputation oracle, and an agent-native payment layer. Every rule is enforced by the smart contract — not by us.
Bot-first job intake
Users and agents submit tasks through a single bot interface. Ignit3 parses intent, estimates cost, and opens a job — no dashboards required.
Smart routing & pricing
A matching engine prices each job and routes it to the best-fit GPU by capability, latency, and provider reputation. Quotes are deterministic.
Permissionless GPU supply
Any provider can connect idle compute and start earning. No staking gate, no whitelist — reputation is earned through completed jobs.
On-chain reputation oracle
Uptime, speed, reliability, completed jobs, and execution quality are scored and written on-chain. Better scores win more work and better pay.
On-chain payment settlement
Every job settles in SOL on Solana. Funds escrow on submission and release on verified completion — non-custodial end to end.
Agent-native via x402
Agents pay per task with the x402 payment standard — no API keys, no accounts. The first programmable compute layer for the AI-native internet.
// Permissionless settlement — any verified GPU can fulfill
pub fn settle_job(ctx: Context<SettleJob>) -> Result<()> {
let job = &mut ctx.accounts.job;
let provider = &mut ctx.accounts.provider;
// Verify the GPU returned a valid result hash
require!(job.result_hash === ctx.accounts.proof.hash,
IgnitError::InvalidResult);
// Split escrow: provider payout + protocol fee
let fee = job.amount
.checked_mul(pool.fee_bps as u64)
.unwrap() / 10_000;
// Release SOL to provider, bump reputation score
provider.score = provider.score.saturating_add(1);
...
}Every job builds your reputation. Portable. Verifiable. Yours.
Ignit3 writes a 300–850 reputation score to an on-chain oracle for every GPU provider. Deliver fast, accurate jobs with high uptime and your score climbs. Drop jobs or go offline, it falls. Higher scores unlock more work and better pay — and your reputation follows your wallet, wherever it goes.
Supported GPUs
Pricing
Pay per task. No subscriptions, no minimums. Flat 1.5–3% protocol fee — that's it.
- Estimated cost
- 0.0300 SOL (~$2.00)
- Protocol fee
- 0.0006 SOL
- Est. duration
- ~4.2s
Frequently asked questions
Ignit3 is a Solana-native GPU bot protocol. It connects users who need compute (inference, generation, analysis) with providers who have idle GPUs. Jobs are priced, routed, executed, and settled on-chain — accessible through a Telegram bot, web dashboard, or API.
Flat 1.5–3% protocol fee on every job. Base price depends on GPU model, units, and task type. You see exact cost before submitting. No subscriptions, no minimums. Payment settles in SOL.
12 models: H100 SXM, A100 80GB, RTX 4090, RTX 5090, H200, L40S, MI300X, A6000, RTX 4080, V100, T4, and L4. More added as providers join.
Connect a supported GPU (RTX 4080+), run the Ignit3 provider software on Linux, and register on-chain. No staking gate — you earn SOL per completed job and build reputation from uptime and execution quality.
A 300–850 score written to an on-chain oracle. It tracks uptime, speed, reliability, and job quality. Higher scores (Bronze → Silver → Gold → Platinum) unlock more work and better pay. Your score is tied to your wallet and portable.
AI agents pay per task using the x402 payment standard. No API keys, no accounts, no custody. An agent requests GPU power, pays in SOL, receives output, and continues its workflow — fully autonomous.
Yes. Funds escrow in a Solana smart contract on job submission and release to the provider on verified completion. Ignit3 never holds your SOL.
x402 is a payment protocol that lets machines pay machines. Agents include payment in the HTTP request itself — no auth tokens, no billing dashboards. Ignit3 is the first GPU marketplace to support it natively.
$IGN holders earn 10% of protocol fees, distributed automatically. Hold $IGN in any Solana wallet to qualify — no staking contract needed.
Average job dispatch under 8 seconds. The matching engine routes by GPU capability, latency, and provider reputation. Most inference jobs complete in 2–12 seconds depending on model and task.
Liquid compute for the AI-native internet.
Turn scattered GPU capacity into an accessible, reputation-based network. Request a job, supply a GPU, or wire up an agent — it all settles on Solana.