Endpoints
A load-balanced reverse proxy for Bittensor RPC, plus the registry of callable Subtensor and subnet endpoints behind it.
One endpoint for Bittensor RPC
POST JSON-RPC to a single URL and Metagraphed routes it across the healthiest, most in-sync public endpoints — with failover, edge caching, and abuse controls. No key, no account, no single point of failure.
https://api.metagraph.sh/rpc/v1/finneycurl -s 'https://api.metagraph.sh/rpc/v1/finney' \
-X POST -H 'content-type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"chain_getHeader","params":[]}'Health-aware load balancing
Requests fan out across the public endpoint pool, weighted by live probe health and score.
Block-height routing
Nodes trailing the freshest reported tip are demoted behind synced peers, so reads hit current state.
Automatic failover
A dead or transient upstream is retried against the next-best endpoint — transparent to the caller.
Edge caching
Block-pinned reads (chain_getBlockHash/Block/Header) are cached at the edge for instant repeat hits.
Read-only + rate-limited
Only safe read methods pass the allowlist; 100 requests/min per client keeps the pool healthy.
Proxy usage
Updated
Snapshot freshness unknown — verify before relying on this data.
| Endpoint | Provider | Requests | Share | Errors | Avg latency |
|---|---|---|---|---|---|
| metagraphed-fullnode-rpc | metagraphed | 13 | 21.0% | 0.0% | 211 ms |
| onfinality-finney-rpc | onfinality | 13 | 21.0% | 0.0% | 260 ms |
| opentensor-lite-rpc | opentensor | 13 | 21.0% | 0.0% | 313 ms |
| opentensor-finney-rpc | opentensor | 10 | 16.1% | 0.0% | 400 ms |
| opentensor-archive-rpc | opentensor | 9 | 14.5% | 0.0% | 517 ms |