- 54%
- Accuracy
- 10 of 14
- Analyst ranking
- 8
- Predictions scored
- 3
- Still to fall due
Artificial Intelligence ▾
Mostly wrong · 25%I predict that by July you will have full access to models that are at least as good as Mythos for code analysis.
What happened. Anthropic kept Claude Mythos Preview restricted and announced Claude Mythos for general use on 15 September 2026, about six weeks after the deadline. This pass found no widely available model shown to match Mythos on code analysis by July. Full access to Mythos-class capability arrived, but after the deadline. Whether another model was 'at least as good' by July is arguable, so confidence is low. Sources: anthropic.com.
Source: More Mythos and Mozilla (The Security Industry, Substack) · 'by July', read generously as the end of July
Wrong · 0%GPT5 is going to be ten times as disruptive and it will likely be released this year.
What happened. OpenAI released GPT-5 on 7 August 2025, not in 2024. The release came about eight months after the deadline. Only the release-date part is scored, since 'ten times as disruptive' cannot be measured. Sources: en.wikipedia.org.
Source: You Ain't Seen Nothin' Yet (The Security Industry, Substack)