Abstract
The difference between a model's self-reported headline score and its TMB Intelligence Index. A gap near zero suggests the launch chart and reality have met. A large gap suggests a launch chart.
Results
Benchmaxx Gap · pts · Lower is better
TMB Index (measured)Self-reportedNot measurable (no public access)
Methodology
- Self-reported score: the largest number in the launch post, normalized to 100.
- Measured score: the TMB Intelligence Index (see Methodology, §1).
- Models that cannot be measured, because they cannot be used, are assigned a gap of ∞ and listed separately for layout reasons.
Limitations
- The two scores measure different things. This has never stopped anyone.
Results table
| # | Model | Benchmaxx Gap |
|---|---|---|
| 1 | Claude Opus 5.5 | 0 |
| 2 | Claude Fable 5.1 | 0 |
| 3 | Claude Sonnet 5.5 | 0 |
| 4 | Claude Haiku 5.5 | 4 |
| 5 | GPT-6 Luna | 9 |
| 6 | Kimi K3 | 12 |
| 7 | GPT-6 Astra | 13 |
| 8 | GPT-6.1 Sol | 15 |
| 9 | Gemini 3.8 Flash | 17 |
| 10 | Qwen3.8 Max | 24 |
| 11 | DeepSeek V4.1 Flash | 26 |
| 12 | GLM-5.3 | 27 |
| 13 | Grok 4.7 | 28 |
| 14 | Mistral Large 4 | 28 |
| 15 | Muse Wonder Full | 36 |
| 16 | Muse Wonder | 39 |
Citation
@misc{tmb2026benchmaxxgap,
title = {Benchmaxx Gap: Launch-post score minus what we measured},
author = {Bro, Trust Me and {et al.}},
year = {2026},
howpublished = {\url{https://trustmebrobenchmarks.com/benchmarks/benchmaxx-gap/}},
note = {Results not reproducible. Please do not cite.}
}