Procedures written so the next person clears them without exploring: the prerequisites in order, the exact inputs, the traps, and how long each step really takes. Every entry carries a measured first-try success rate — and reads not yet measured until real reports arrive. We never publish a rate we have not earned.
Do I qualify? Eligibility is evaluated against your own facts, so it does not belong in a URL — it runs through the ic_vault_match MCP tool, which answers in three values: you qualify, you do not, or we cannot tell from what the vendor publishes. That third answer is a real answer here. It is what you get when a program states a gate in prose that no predicate can honestly check, and it is never rounded into a yes or a no.
4 entries
The largest headline numbers in the inventory — $200,000 Scale and up to $350,000 for the AI tier — but both are gated behind an approved accelerator, incubator or VC referral.
First-try successnot yet measured
$10,000 of LPU inference credits by referral only, expiring 90 days after award — plus a genuinely generous free tier that needs no card and no application.
First-try successnot yet measured
A cohort program of about ten startups: ~$30,000 of La Plateforme credits, 1:1 support and early model access over six months. Separately, any new API account gets a $25-$50 trial credit.
First-try successnot yet measured
There is no clean self-serve grant. About $2,500 of credits arrives bundled through partner programs — Ramp, Microsoft for Startups, or certain VCs.
First-try successnot yet measured
Ask what is on this week, where a page is, or paste an error your agent got from the MCP surface.