AUTO
You ask. Ada routes.
Ada scores the request and chooses the appropriate tier automatically.
score 18FAST
score 47SMART
score 82MAX
DIFFICULTY-BASED BY DESIGN
ADA / COMPUTE CORE
REQUEST RECEIVED
DIFFICULTY72
FASTSMARTMAX ↗
ADA COMPUTE CORE / CONCEPT VISUAL
ONCHAIN COMPUTE. OFFCHAIN INTELLIGENCE.
Burn $ADA.Get compute.One API for AI.
Fund compute with $ADA. Access AI through one Ada key. Let model: "auto" match every request to the compute it needs.
01Intelligent routing
The right compute for the work.
Not every request needs the same engine. Ada is designed to score each model: "auto" request before inference, then route it to the right tier.
ONE MODEL PARAMETER.
THREE COMPUTE TIERS.
THREE COMPUTE TIERS.
FAST
Less wait. Less compute.
Low latency. Lower compute cost. For work that doesn't need to overthink.
Rewrites & summariesData extractionSimple lookupsTransformations
LIGHTWEIGHT WORK
SMART
Your everyday engine.
Substantial compute for the coding, reasoning and analysis you do every day.
Everyday codingGeneral reasoningStructured analysisDeveloper tasks
SUBSTANTIAL WORK
MAX
Bring the hard problems.
Premium compute for the requests that need depth, context and careful thought.
ArchitectureDifficult debuggingLong-context analysisComplex reasoning
DEMANDING WORK
Routing concept only. Scores are illustrative; no live inference or routing is connected.
02The compute economy
Compute that grows from the network.
A direct path from $ADA to AI. With a compute pool designed to make the network's activity work for its users.
Base compute
Compute obtained directly from burning $ADA. Your starting point for the work ahead.
Bonus compute
Creator fees can fund the Ada Compute Pool, providing additional compute when funded and available.
01 / THE DIRECT PATH
$ADA
BURN
BASE COMPUTE
ADA API
02 / THE NETWORK LOOP · WHEN FUNDED
$ADA TRADING
CREATOR FEES
COMPUTE POOL
BONUS COMPUTE
Base compute10,000
Bonus compute2,000
Total compute12,000
ILLUSTRATIVE EXAMPLE · NOT A LIVE RATE OR GUARANTEED BONUS
03Built for developers
One endpoint. One key.
An OpenAI-compatible experience, designed to fit your existing workflow. The compute decisions stay with Ada.
- 01Burn $ADA
- 02Receive compute
- 03Create an Ada key
- 04Send
model: "auto" - 05Ada routes the request
Planned workflow. Product functionality is not live.
Example request
cURL
curl https://api.adacompute.fun/v1/chat/completions \
-H "Authorization: Bearer ada_sk_..." \
-H "Content-Type: application/json" \
-d '{
"model": "auto",
"messages": [
{
"role": "user",
"content": "Explain this code."
}
]
}'
04A few answers
Less mystery. More clarity.
The essentials behind Ada, its compute economy and the developer experience.
Put your compute to work.
Explore the Ada console. The foundation is here.