Use Inference Pool
A chat app and API over many models. Send a prompt, the router picks eligible GPU supply, and tokens stream back in real time.
Open the appOne pool
for GPU inference.
Every GPU on the network is one pool of supply. Send a request, the router draws from it, and tokens stream back. A Uniswap v4 hook pays the same pool from $INFP liquidity.
Settles onRobinhood Chain· $INFP
App · APIRouter
GPU workerToken stream
Pooled
Every GPU, one pool of supply
Streamed
Token-by-token response
Permissionless
Any GPU can join the pool
0.30%
Of every swap, into the reward pool
How the pool routes a request.
The router draws from the pool by model, availability, and measured speed. The worker executes and tokens stream straight back. Deeper the pool, faster the match.
Worker benchmarkBlackwell-class reference
pool-open-8bfirst token 0.24s
168 tok/s
pool-mistralfirst token 0.21s
182 tok/s
pool-reason-r1first token 0.38s
96 tok/s
pool-reason-minifirst token 0.14s
340 tok/s
Every worker runs inferencepool worker benchmark on join and reports its own numbers. The router weights routes by measured tokens per second, not by advertised hardware.
NVIDIA · Blackwell
GB202 / GeForce RTX 5090
21,760
CUDA cores
6805th gen
Tensor cores
1704th gen
RT cores
32 GBGDDR7
Memory
512-bit
Bus width
92.2B
Transistors
4 nmTSMC 4N
Process
575 W
TGP
Memory bandwidth by generation
RTX 3090
936 GB/s
RTX 4090
1008 GB/s
RTX 5090
1792 GB/s
Specs from NVIDIA's RTX Blackwell architecture whitepaper. Inference Pool workers report the same fields, VRAM, cores, bandwidth, for routing.
Usage is counted, earnings are credited, and prompt context is discarded once the response completes.
Two pools, one reward pool.
Compute liquidity and token liquidity feed the same place. Holders earn from fees that were actually collected, never from emissions.
The compute pool
Every GPU, one pool
Workers join and advertise the models they serve and how fast they serve them. The router draws from that pool per request. Supply is pooled, not assigned, so a deeper pool means a faster match and better fan-out.
The liquidity pool
Swaps pay the same pool
$INFP trades in a Uniswap v4 pool with a custom hook attached. On every swap the hook takes 0.30% and routes it straight into the holder reward pool that inference margin already fills. Capped at 1% in the contract, and renounceable.
Two sides of the network.
Provide compute
Run the worker client, benchmark your machine, list supported models, and accept routed jobs. Earnings are credited per usage.
Become a providerStake and earn
Holders stake for a pro-rata share of the reward pool. It fills from two directions: 80% of every job's margin, and 0.30% of every $INFP swap through the Uniswap v4 hook.
Read tokenomics