Abstract
Calendar days between the first public claim about a model and the first day a member of the public could send it a prompt. Open intervals are measured as of the build date of this page, and keep growing.
Results
Days to Ship · days · Lower is better
Methodology
- Start: the earliest of announcement, teaser, or leadership tweet.
- End: general availability via API or app. Waitlists, "trusted partners", and screenshots do not count.
Limitations
- Some announcements are jokes. We cannot always tell which. Neither, it seems, can the labs.
Results table
| # | Model | Days to Ship |
|---|---|---|
| 1 | Claude Opus 5.5 | 0d |
| 2 | Claude Fable 5.1 | 0d |
| 3 | Claude Sonnet 5.5 | 0d |
| 4 | Claude Haiku 5.5 | 0d |
| 5 | GPT-6 Astra | 0d |
| 6 | GPT-6.1 Sol | 0d |
| 7 | GPT-6 Luna | 0d |
| 8 | Gemini 3.8 Flash | 0d |
| 9 | Muse Wonder Full | 0d |
| 10 | Muse Wonder | 0d |
| 11 | Grok 4.7 | 0d |
| 12 | Kimi K3 | 0d |
| 13 | GLM-5.3 | 0d |
| 14 | Qwen3.8 Max | 0d |
| 15 | DeepSeek V4.1 Flash | 0d |
| 16 | Our Model™claimed | 1d |
| 17 | Claude Nextclaimed | 2d |
| 18 | Gemini 4 Argonclaimed | 13d |
| 19 | Mistral Large 4 | 113d |
| 20 | Le Chaton Fatclaimed | 121d |
| 21 | Grok 5claimed | 419d |
| 22 | Reflection 405Bclaimed | 769d |
Citation
@misc{tmb2026shiplatency,
title = {Announce-to-Availability Latency: Days between "introducing" and "available"},
author = {Bro, Trust Me and {et al.}},
year = {2026},
howpublished = {\url{https://trustmebrobenchmarks.com/benchmarks/ship-latency/}},
note = {Results not reproducible. Please do not cite.}
}