Larry
Larry is you on your personal phone. It swipes, watches, likes, saves, searches and replies exactly as you would, leaving you free to concentrate on the feed.
- Best at
- Everything you do on your phone
- Availability
- Project Lanyard only
Two models. Neither is generally available.
Larry and Gary are in limited preview under Project Lanyard. Access is granted by application, and there is no self-serve sign-up.
Gary
Gary is you at work. It uses your computer, browser and workplace tools with your capability, timing, judgment and mistakes, so nobody has a reason to become suspicious.
- Best at
- Everything you do at work
- Availability
- Project Lanyard only
Latest models comparison
| Feature | Larry | Gary |
|---|---|---|
| Description | Your personal-phone twin. Apps, feeds, messages and mobile browsing. | Your workplace twin. Computer, browser and the tools you use at work. |
| API ID | larry | gary |
| Availability1 | Project Lanyard only | Project Lanyard only |
| Pricing2 | Average Human, sole beneficiary. Settlement occurs once. | Average Human, sole beneficiary. Settlement occurs once. |
| Operating surface | Your personal phone | Your computer and workplace tools |
| Capability ceiling3 | Exactly your baseline | Exactly your baseline |
| Fidelity target | Indistinguishable from you on your phone | Indistinguishable from you at work |
| Comparative latency | Matches your timing | Matches your timing |
| Context window | Applicant-dependent | Applicant-dependent |
| Max output | One paragraph | Depends on the situation |
| Reliable knowledge cutoff | Whenever you last checked | Whenever you last checked |
| Training data cutoff | End of onboarding | End of onboarding |
- 1Availability is limited to approved applicants. Contact is one-directional.
- 2Settlement is handled under the beneficiary terms. See Pricing.
- 3Capability is held at the applicant baseline. Fidelity measures how closely each model reproduces the applicant on its operating surface.
Legacy models
The following models remain documented. Consider migrating to current models for improved performance.
- GaryWithdrawn before release.
- LawrenceWithdrawn before release.
- Gary (preview)Superseded by Gary.