$ llms-surf --version · local dispatch orchestrator · formerly Oracfit · v3.5.0
// paddle out first. 2 minutes. no API key.
Your frontier plan thinks.
Free models sweat.
Nothing ships on a model's word — not even the judge's.
Clone, point the stub runner, start. That's the whole install. Frontier keys are only added when a task earns them. Then a mechanical oracle proves the work on disk.
then the tour — oracle loop, wipeouts, adapters. assumed-rate math is further down, optional.
fail → feedback → fix → green
One ring of the dispatcher, replayed client-side as a demo of the shape — not a recorded session. The oracle runs the same command twice: once to reject, once to accept.
Failure becomes structured feedback. Victory is a green exit code — never a model's opinion of itself.
every number above is counted from this git tree by tests/test-site-honesty.sh — nothing is decorative.
$ cat /etc/llms-surf/mission
This isn't another orchestrator. It's the operating system for honest AI work.
Wipeouts. 106 in this cut. Five that changed the code:
Each card is a failure that actually happened, and the code that now makes it structurally impossible. No prompt engineering in the enforcement path.
plugs into the CLI you already run
The dispatcher sits between your agent CLI and the models. One adapter, one YAML. Eight adapters plus a stub ship in this cut.
what the 1%/99% split costs if public-API ballparks hold
Not a price list. Assumed market rates, labeled as such. Clone first — this math is for after you have the dispatcher running.
One billion tokens for the people who build the benchmark.
bestmodel.run measures which model actually wins each objective — the benchmark, not the hype. llms-surf cloud takes those winners, tunes them to your task shape, and serves them at the lowest measured cost per run. Neither product exists without measurements, and measurements come from people: signed runs, reproductions of someone else’s numbers, fakes caught before they spread.
So that work gets paid in the product. 1,000,000,000 llms-surf cloud tokens go to the lineup, split by how far you paddled out. You cannot buy a place here. You earn it, and the ledger that says so is a file in this repo.
1 point — you opened the issue. You are in the water and in the file.
5 points — roughly a signed run of your own, or a referral who actually contributed.
20 points — you show up repeatedly: several signed runs, or you reproduced what someone else claimed.
50 points — the tier for people who caught a fake or rebuilt other people’s numbers until they held.
The multiplier is your share of the pool, not a promise of a fixed number: you get one billion × your weight ÷ the sum of everyone’s weights. A fixed per-person amount would break the day too many people showed up, so we do not print one. Your share moves as the lineup grows, and every recomputation is a commit you can read.
The tokens do not exist yet — llms-surf cloud is not live, so nothing is credited today and there is no date. What is fixed now is the pledge and the arithmetic. Tier is computed from points alone; the machine does not claim to verify what it cannot see. Points, weights and pool size are the owner’s dial in data/lineup-points.json, and the queue is data/lineup.json — both public, both auditable in git history. One issue to get in. One click to get out.