
Overview
Registration Activity
Neuron registration trends

Neuron registration trends
The alpha price in τ, day by day, and what moved it each way.
Log change over the last 30 days −39.14%: protocol buys +21.59%, root sells −2.05%, users net −60.73%, cross-subnet +2.05%, unattributed −0.0001%.
The chart follows the alpha price in τ. Each period starts at its opening price and steps through the mechanisms below, in a fixed order, to its closing price: a step that rises pushed the price up, one that falls pushed it down, and its height is how far it moved the price. A thin line carries each close into the next period’s open; it is dashed where the two differ. The faint candle behind each period is the market’s own open, high, low and close, for context.
With many periods on screen (90 days, or hours), the steps would be too thin to read, so the chart draws the price as a line instead and colours each period by the mechanism that moved it most. Hover a period for the full step-by-step breakdown.
Not here, because they do not move the spot price: the per-block TAO injection into the pool, and burns.
All figures are log changes, not simple percentage changes, so the parts add up exactly to the total; on big moves the two differ (a 10% log fall is a 9.5% price fall).
2026-11-11 UTC, as of ≈14:38 UTC (block 9,483,678)
The day isn’t over yet, so these figures keep changing. They are in no daily total, average or rank.
Holder & circulating-alpha metrics as of (daily close)
Running frontier open models at scale requires trusting inference providers to serve the model they claim—not a cheaper quantization or distillation masquerading as the original. Traditional centralized inference hides this risk behind a brand; decentralized networks of untrusted GPUs cannot. engy solves this by attaching a cryptographic commitment (TOPLOC proof) to every response, verifying that the model's activations match the canonical checkpoint. No TEEs or trusted hardware—the proof pins the math, not the machine.
A permissionless network of GPU miners serves frontier open models and proves compliance to validators who sample requests and recompute the proofs. The architecture splits proving (zero extra GPU cost, async tee off the response path) from verification (sampled, out-of-band). Because proof capture is fire-and-forget and verification runs independently, slow or dead validators never degrade serving—the SLA is owned by the gateway's admission gate, not consensus latency.
SN53 is live today with Qwen3.6-35B-A3B (35B/3B-active MoE, FP8) on consumer hardware—2× RTX 4090, single L40S, or datacenter cards. Miners register permissionlessly after proving they serve the canonical model through probe traffic. Light validators (CPU-only, no GPU, no model weights) fetch the master-signed epoch result, verify the signature, renormalize hotkeys against the chain metagraph, and submit on-chain weights every ~100 blocks to stay inside activity_cutoff. Scoring is per-miner, per-epoch, gated on acceptance rate, latency, and audit verdicts—only billed traffic mints weight.
Auto-generated from the GitHub repository and may not reflect the latest changes.
engy (SN53) is a Bittensor subnet providing verifiable inference for frontier open models on consumer GPUs. Miners serve committed model checkpoints with cryptographic TOPLOC proofs attached to every response; validators verify samples and set on-chain weights. Currently live with Qwen3.6-35B-A3B (35B/3B-active MoE, FP8) on 2× RTX 4090 or better hardware, verified end-to-end against published Merkle roots.
Buyers connect to a single OpenAI/Anthropic-compatible endpoint at api.engy.ai. The gateway authenticates, handles billing, and load-balances across 1st-party clusters and permissionless subnet miners equally via round-robin with prefix affinity. Every response carries a TOPLOC proof.
New subnet miners are probed on synthetic traffic before serving buyers; the gateway pulls any that fail SLA. Miners dial out over websocket to the gateway (no inbound port); they tokenize locally, call sglang serve, read hidden states, build the proof, and reply with completion+proof. The master validator (GPU-backed, engy-run) samples requests, recomputes TOPLOC proofs, scores each miner, and publishes a signed epoch result.
Light validators (any third party, CPU-only) fetch the signed result, verify the signature against the pinned master hotkey, renormalize and cross-check hotkeys against the chain metagraph, then submit on-chain weights via Yuma consensus every ~100 blocks to stay active. The chain distributes emissions.
Per-(miner, model) pair, per epoch (weekly in prod): score = floor(Σ tokens(r) × score_rate[model] / 1000) if all four gates pass, else 0. Gates: (1) acceptance >99%, (2) TTFT p99 below model spec, (3) TPOT p99 below model spec, (4) cheat verdicts <1% of samples. Each gate has a minimum sample size below which it passes automatically.
Gates are applied in order; first failure zeroes the epoch. 'unproven' verdicts never gate. Scores renormalize to 65535; if zero total, the full amount goes to the owner hotkey (burn). On-chain weights apply under Yuma consensus.
The sections above were auto-generated from the GitHub repository and may not reflect the latest changes. View repository (github.com, opens in a new tab, external site)
Verified inference for frontier open models.
This information was auto-extracted from the GitHub repository and may be incomplete or out of date. Always refer to the subnet's GitHub repository (github.com, opens in a new tab, external site) for the latest instructions.
Not yet verified: GPU, Notes
Not yet verified: Network, Notes
Machine-extracted from the subnet's repository docs; may be wrong or out of date. Check the repository before buying or renting hardware. Open the repository (github.com, opens in a new tab, external site) · Extracted 5 Oct 2026
Install sglang model server and miner dependencies
pip install "sglang>=0.4.6" toploc transformers torch numpy requests websockets
Copy the miner module (single file) to deployment location
cp miner/engy_miner.py /path/to/miner/
Download and place the Qwen3.6-35B-A3B FP8 checkpoint locally
# Place model checkpoint on local disk, e.g. /data/models/Qwen/Qwen3.6-35B-A3B-FP8
Register and obtain a miner key from the provider dashboard
# Obtain miner key from https://provider.engy.ai
Start sglang serve with hidden-states and parser support (reference config for 4× RTX 4090)
PYTHONPATH=/path/to/engy/miner python -m sglang.launch_server --model-path /data/models/Qwen/Qwen3.6-35B-A3B-FP8 --served-model-name Qwen3.6 --tp-size 4 --trust-remote-code --kv-cache-dtype fp8_e4m3 --mem-fraction-static 0.83 --chunked-prefill-size 8192 --max-running-requests 8 --context-length 262144 --enable-return-hidden-states --enable-cache-report --tool-call-parser qwen3_coder --reasoning-parser qwen3 --host 0.0.0.0 --port 8000
Verify sglang server is running and responding
curl -s http://127.0.0.1:8000/get_model_info >/dev/null && echo "serve up"
Start the miner (set MAX_INFLIGHT to serve concurrency; minimum 8)
GW=wss://lb.engy.ai/gw MINER_KEY=<your-key> MODEL=qwen3.6-35b-a3b MAX_INFLIGHT=8 python miner/engy_miner.py --checkpoint /data/models/Qwen/Qwen3.6-35B-A3B-FP8 --serve-url http://127.0.0.1:8000
Clone the engy repository
git clone https://github.com/hanlinai/engy && cd engy
Copy the validator environment template
cp .env.validator.example .env.validator
Fill in ENGY_SN53_WALLET and ENGY_SN53_WALLET_HOTKEY from btcli wallet list
# Edit .env.validator with your wallet name and hotkey name
Start the validator container with Docker (includes auto-update via Watchtower)
docker compose --env-file .env.validator -f docker/docker-compose.validator.yml up -d
Monitor validator logs to confirm operation
docker compose -f docker/docker-compose.validator.yml logs -f validator
Confirm wallet configuration in logs and validate on-chain registration
# Verify startup echoes wallet and hotkey confirmation, then check btcli for current weights and last_update
| Day (UTC) | Share |
|---|---|
| 2026-11-10 | 3.59% |
| 2026-11-09 | 3.80% |
| 2026-11-08 | 3.96% |
| 2026-11-07 | 3.94% |
| 2026-11-06 | 3.77% |
| 2026-11-05 | 3.94% |
| 2026-11-04 | 4.16% |
| 2026-11-03 | 4.26% |
| 2026-11-02 | 4.23% |
| 2026-11-01 | 4.27% |
| 2026-10-31 | 4.30% |
| 2026-10-30 | 4.24% |
| 2026-10-29 | 4.40% |
| 2026-10-28 | 4.28% |
| 2026-10-27 | 4.21% |
| 2026-10-26 | 4.29% |
| 2026-10-25 | 4.22% |
| 2026-10-24 | 4.25% |
| 2026-10-23 | 4.16% |
| 2026-10-22 | 4.20% |
| 2026-10-21 | 4.46% |
| 2026-10-20 | 4.57% |
| 2026-10-19 | 4.68% |
| 2026-10-18 | 4.79% |
| 2026-10-17 | 4.90% |
| 2026-10-16 | 5.07% |
| 2026-10-15 | 5.19% |
| 2026-10-14 | 5.19% |
| 2026-10-13 | 5.04% |
| 2026-10-12 | 5.00% |
| 2026-10-11 | 4.86% |
| 2026-10-10 | 4.69% |
| 2026-10-09 | 4.58% |
| 2026-10-08 | 4.75% |
| 2026-10-07 | 4.69% |
| 2026-10-06 | 4.45% |
| 2026-10-05 | 4.31% |
| 2026-10-04 | 4.36% |
| 2026-10-03 | 4.65% |
| 2026-10-02 | 4.87% |
| 2026-10-01 | 4.68% |
| 2026-09-30 | 4.66% |
| 2026-09-29 | 4.25% |
| 2026-09-28 | 4.32% |
| 2026-09-27 | 4.63% |
| 2026-09-26 | 4.98% |
| 2026-09-25 | 4.87% |
| 2026-09-24 | 4.61% |
| 2026-09-23 | 4.75% |
| 2026-09-22 | 4.30% |
| 2026-09-21 | 3.95% |
| 2026-09-20 | 3.84% |
| 2026-09-19 | 3.64% |
| 2026-09-18 | 3.53% |
| 2026-09-17 | 3.60% |
| 2026-09-16 | 3.58% |
| 2026-09-15 | 3.59% |
| 2026-09-14 | 3.72% |
| 2026-09-13 | 3.38% |
| 2026-09-12 | 3.25% |
| 2026-09-11 | 3.06% |
| 2026-09-10 | 3.07% |
| 2026-09-09 | 3.36% |
| 2026-09-08 | 3.69% |
| 2026-09-07 | 4.15% |
| 2026-09-06 | 4.59% |
| 2026-09-05 | 4.72% |
| 2026-09-04 | 4.90% |
| 2026-09-03 | 4.96% |
| 2026-09-02 | 4.54% |
| 2026-09-01 | 3.89% |
| 2026-08-31 | 4.00% |
| 2026-08-30 | 4.03% |
| 2026-08-29 | 4.07% |
| 2026-08-28 | 4.26% |
| 2026-08-27 | 3.89% |
| 2026-08-26 | 3.76% |
| 2026-08-25 | 3.78% |
| 2026-08-24 | 3.92% |
| 2026-08-23 | 3.90% |
| 2026-08-22 | 4.06% |
| 2026-08-21 | 4.34% |
| 2026-08-20 | 4.34% |
| 2026-08-19 | 4.47% |
| 2026-08-18 | 4.58% |
| 2026-08-17 | 4.60% |
| 2026-08-16 | 4.56% |
| 2026-08-15 | 4.26% |
| 2026-08-14 | 4.32% |
| 2026-08-13 | 4.33% |