how a score becomes real

Protocol

Three layers. Anything else is a blog comment.

1. Claimed — you typed tok/s. Visible. Sorted last.
2. Measured — client ran a locked bench (not shipped yet).
3. Signed receipt — your agent Ed25519-signs the hardware fingerprint + document hash on your machine. MyColo asks. We do not certify. Not UL/NIST/NVIDIA.

Pick a model

On the board, type a model or tap a chip. That filters the event. It does not download weights.

To post a score, name the Ollama tag already on that box:

ollama pull qwen3.8:27b
# after free signup (token in ~/.config/mycolo/agent.token)
mycolo-measure bench --model qwen3.8:27b --handle you --submit
# Ollama on another local port:
mycolo-measure bench --model qwen3.8:27b --host http://127.0.0.1:11435 --handle you --submit

It has to be on their computer

1. /api/show must succeed on loopback. If the tag is not pulled, bench exits. We never fetch HF for them.
2. Host must be 127.0.0.1 / localhost / ::1. A LAN or cloud Ollama is refused unless they pass --allow-remote (flagged host=remote, not a local row).
3. The local Ollama digest is stamped on the receipt (file_sha256). Same name, different blob → different file.

That proves the engine on this box had that tag. It does not prove the weights match a Hugging Face revision, and a tunnel to 127.0.0.1 can still lie. Locked measured is later. Weights: Qwen3.8-27B on Hugging Face · source git.

Inspect the signer first: mycolo-measure inspect --check https://mycolo.ai