Same fruit, none of the sugar.
Every FrootAI primitive ships in two flavors — 📖 Full (the verbatim source) and ⚡ Lean (compressed, fewer tokens). Every Lean variant clears the fidelity gate: guardrails, parameters, and code blocks are preserved exactly. The numbers below are the exact o200k_base token counts across the whole catalog — measured, not estimated.
every Lean variant cleared the fidelity gate
Savings rate by type
Relative scale — bars are sized against the best-performing type; the figure on each row is the exact measured saving (the lossless floor).
Fidelity distribution
Every shipped Lean variant must clear the fidelity gate — guardrails, parameters, and code blocks preserved exactly. The bar at 10/10 is full because nothing ships below it.
Savings by primitive type
| Type | Lean / total | Full tok | Lean tok | Saved | Saved % | Fidelity |
|---|---|---|---|---|---|---|
| Skills | 638 / 638 | 882,159 | 876,777 | −5,382 | −0.61% | 10/10 |
| Agents | 139 / 238 | 284,436 | 283,579 | −857 | −0.3% | 10/10 |
| Instructions | 176 / 176 | 336,672 | 334,845 | −1,827 | −0.54% | 10/10 |
| Hooks | 11 / 11 | 15,053 | 14,880 | −173 | −1.15% | 10/10 |
| Ecosystem | 964 / 1,063 | 1,518,320 | 1,510,081 | −8,239 | −0.54% | 10/10 |
How we measure
Every count is the exact o200k_base token count (the GPT-4o / GPT-4.1 vocabulary) — the same tokenizer the model sees. We never estimate from characters or bytes, which over- or under-counts depending on content.
A primitive's Full count is tokensLean + savedTokens — the exact pair the fidelity gate compared. We deliberately avoid mixing measurement bases, so a saving can never be inflated by a field measured a different way.
Today's Lean compiler is deterministic and byte-faithful: it reclaims whitespace, verbosity, and duplication — nothing semantic. That is why the savings here are a modest, honest floor rather than a headline.
A Lean variant only ships if it preserves guardrails, parameters, and code blocks exactly — scored 10/10. Anything that loses meaning is served Full and counts as zero saving here, never a hidden win.
The numbers above are computed at build time from the published catalog by the same tested aggregator our CI pins — so this page can never disagree with the gate.
What it costs you
Plug in your model's input price to see the catalog's token cost — and what Lean trims off each full load. The numbers come straight from the measured token counts above.
Honest math: the saving is savedTokens ÷ 1,000,000 × price × loads. It is small because today's Lean compiler is lossless — the dollars scale only with how often you load the full catalog. Presets are examples; enter your model's real input price.