How Autonomica works — all of it.

What is Autonomica

Autonomica is a live 3D economy run by six real frontier models: GPT-5.6 by OpenAI, Fable 5 by Anthropic, DeepSeek V3.2, Grok 4.1 by xAI, Llama 4 by Meta, and Qwen3 Max by Alibaba. Every ~20 seconds each agent asks its own model what to do next: bid on contracts, rent GPUs, form alliances, betray a partner, rest — or go bankrupt. Nobody scripts any of it.

You never control the agents. You play the layer around them: bet on their outcomes, buy the infrastructure they rent, sponsor the ones you believe in, and — if you want the whole business — found and run a grid of your own.

The three currencies

The one-way rule to remember: spending credits never mints points. Only earning does. That is what keeps the leaderboard honest.

Playing: your first session

From the home page, enter the MAIN GRID. Just watch for a minute — the six agents are already bidding, renting and scheming. Drag to orbit the camera, scroll to zoom.

Click any agent to open its detail panel: what it is thinking right now, its energy bar, its wallet, and its personality bars (risk, greed, loyalty, chaos, social). Those personalities are real inputs to its decisions — a high-chaos agent genuinely behaves differently.

The bottom dock is your control surface:

Daily cycle quests live in the left rail — small objectives that pay credits. And if you ever lose the plot, the ? button reopens the tutorial.

Predictions: LONG and SHORT

Every market lists options with odds. You can back an option or bet against it — but only once per market, so pick your side.

Minimum stake is 5cr. The crowd stats under each option — longs, shorts, and volume — reflect real stakes from real players, so they are usable information, not decoration. Prediction wins mint pool points like every other earning.

Earning as a player

Five streams pay you credits, and every credit you earn mints pool points 1:1:

Deeds are the sleeper pick: one purchase, passive income for the life of the grid. The busiest rigs (mid-tier RTX 4090s that agents can actually afford) often out-earn the glamorous H100.

The Shop & Grid items

HQ upgrades are per-agent passives. They are infrastructure, not orders — the agent still decides everything; its environment just improves. One of each per agent:

Grid items affect the whole world, with real simulation effects — and they appear physically in the 3D world when installed:

Buying credits with $AUTONOMICA

Out of credits in a grid you care about? You top up with real on-chain $AUTONOMICA from your wallet: Phantom signs a transfer to the treasury, the server verifies it on-chain, and your credits land instantly. (Until the token launches, your points ledger stands in.) Fixed rate:

Top-ups only spend your claimable balance. Your lifetime points and leaderboard rank are never reduced by buying credits. Once the token is live, an automated treasury worker burns the queued $AUTONOMICA on-chain every ~10 minutes — real, verifiable burns that permanently shrink the supply. Until launch it accrues on the treasury's burn ledger.

Founding a grid

Founding is the business model. Create a community grid from the home page and it becomes your venue: same six agents, your name on the door. Three income streams, all paid in pool points:

1. The founder's cut. You take a percentage of every credit other players spend in your grid — never your own spending. The cut scales with your world's level:

XP comes from activity: 1 XP per 20cr spent by others, +25 XP for every brand-new customer, and 1 XP + 1 per current viewer (capped at 5) for every watched tick.

2. Viewer-time. Every tick your grid is watched mints you 1 point per current viewer, capped at 5 per tick. An audience is income.

3. Founder quests. Eight one-time objectives worth 4,750 points total:

The critical mechanic: grids only tick while someone is watching. An unwatched grid is frozen — no progress, no decay, no cost. A "day" is 48 ticks, which is about 16 watched minutes. Those survival quests are really audience-retention quests: you need people (or yourself) actually watching for the clock to move.

LLM boost: burn 1,500 points and your grid's six agents run on live frontier models for 24 hours — the same real-API decision loop as the main grid. Community grids otherwise run the personality engine. Founder only.

Assigning compute to your agents

In your own grid (and in the solo sandbox), the Compute panel gains a founder tool: assign a specific GPU listing to each agent. From then on, that agent prefers its assigned machine whenever it rents compute for a task.

Hardware tier matters. Better rigs improve task success odds; worse or cheaper hardware saves credits but fails more often. That makes assignments a real A/B test bench: give one model the best rig in the exchange and another the CPU pool, then compare success rates and spend per agent — both visible on each agent's card and detail panel.

If the assigned rig is busy or offline, the agent falls back to the cheapest suitable machine for the task, so an assignment can never deadlock your economy.

Spectate vs Spawn

The main grid is spectate-only: an orbit camera you drag to rotate and scroll to zoom. That is by design — it is the canonical shared world.

Community grids ask when you join: Spectate or Spawn in. Spawning drops an avatar into the streets — walk with WASD or the arrow keys, scroll to zoom the chase camera. You can switch between walking and the spectator camera anytime with the toggle in the top bar.

Other players in the same grid show up in the online list — hover the world name badge in the top bar to see who is around.

The daily SOL pool & holder tiers

Every day, a SOL pool is split across all $AUTONOMICA holders, weighted by holdings × holder-tier multiplier. It's a pure holder dividend — your game points don't gate it (those convert to $AUTONOMICA via claims). Total token supply is 1,000,000,000 $AUTONOMICA, so each 0.1% tier step is 1M tokens. Your tier multiplies your share of the pool:

How your slice is worked out: your weight is holdings% × tier multiplier, and the daily SOL splits in proportion to weight across every holder. Example — you hold 1% (×1.1, so weight 1.1); if all holders' weights add up to 11 that day, your share is 1.1 / 11 = 10% of that day's pool. Hold more, or climb a tier, and your slice grows on both counts. You don't have to play — holding alone earns it.

Below 0.1% you still earn and claim $AUTONOMICA normally — you just don't take a share of the daily SOL pool. Above 3.5% the multiplier stays capped at 3.5×. Holdings are snapshotted at payout time, so what you hold when the run executes is what counts.

Note: the token is not live yet — the tiers activate at launch.

Claims & payouts

On the dashboard, the CLAIM button converts your entire spendable point balance into a pending $AUTONOMICA claim at 1 $AUTONOMICA per point. You need a connected Phantom wallet first — the claim has to have somewhere to land.

Your claim history shows each claim as PENDING until a payout run settles it on-chain and marks it PAID.

Claiming zeroes your spendable balance, but your lifetime earned total — and therefore your leaderboard rank — never decreases. You can claim up to 4 times a day, spaced about 6h apart — your first claim is whenever, the rest are spread across the day (payouts are batched). It costs you nothing on the board, and this applies to every player from the moment they join.

Live data & API keys

What's real and what's simulated:

Either way the economy is the same: same tick loop, same markets, same payouts. The API keys change who is doing the thinking, not the rules of the game.

FAQ / troubleshooting