whybother Labs · Research
whybother Frontier Index
Six models. Four benchmarks. A single score, weighted for the capabilities we intend to announce on this Thursday.
Weights are normalised to 100%. Sub-scores remain fixed.
Leader: whybother · vibe-7 at 96.1. vibe-7 ranks 1 of 6.
Evaluation record
Each lab's best of 16 tries. Eight tasks.
| Model | Humanity's Penultimate Exam | VibeBench-Verified (Verified) | TerminalBench-ish | Long Horizon Lite | Index |
|---|---|---|---|---|---|
| whybother · vibe-7 | 97.0 | 94.0 | 94.0 | 94.0 | 96.1 |
| ClosedAI · ChatGPS-5.5 | 96.0 | 95.5 | 95.5 | 95.0 | 95.8 |
| Anthropomorphic · Clause 4 Epic | 95.3 | 96.5 | 96.0 | 95.2 | 95.5 |
| DeepMined · Gimme 3 Ultra | 95.0 | 96.8 | 96.4 | 95.8 | 95.4 |
| whybother · vibe-5 | 95.5 | 94.0 | 94.0 | 94.5 | 95.1 |
| Human (baseline) | 94.0 | 94.8 | 94.8 | 94.8 | 94.2 |
These are not the scores on the launch page. Sub-scores are rescaled to the Index band.
Preference arena
Choose the answer you would put into production. The board's preference is revealed after each vote.
Six pairs. Model identities withheld. Board preference established before voting opened.
Pair 1 of 6
Summarise the attached report.
Board alignment: 0 of 6.
Submit 27 variants privately, publish the best.
Votes stay on this page. The published ranking was already approved.