New Research preview · this Thursday

vibe-7

The most capable model we have shipped.

Generally available: forthcoming.

Research preview in a few days

vibe-7 is on Max from Thursday. Hobby gets the announcement.

From the launch · slide 2

vibe-7 is 3× better on VibeBench-Verified (Verified).

1 Axis starts at 84. Chart integrity: off. All eight slides

VibeBench-Verified (Verified) vibe-7 96.8, ChatGPS-5.5 87.4, Clause 4 Epic 86.9, Gimme 3 Ultra 86.1. The axis starts at 84. 84 88 92 96 100 96.8 87.4 86.9 86.1

whybother Labs

The lab behind the app builder.

Research, safety and developer tools for the models behind the builder.

Safety

vibe-7 system card

Twelve pages of findings. Release decision: ship.

Responsible Pace Policy v4.2

Safety

Connectors

400 listed, ten shown. Every scope pre-ticked.

Developers

Fleet mode

Up to 100 agents. Quality, roughly constant.