Astra System Card Chart Shows Claude and Gemini Lead on Prompt Injection Robustness
andrew_n_carr · x · 2026-09-04
Andrew Carr shared a chart from the Astra system card testing models' robustness to prompt injections. Key takeaways:
- Claude models perform notably well, with Gemini also strong
- The OpenAI team appears to have made a significant robustness leap between the Sol and Astra generations
- The Luna team's models lag on robustness, which Carr jokes must be stressful
It's one of the first concrete community observations on safety-guardrail gaps across frontier models since the system card's release.
More from Models
- System card data contradicts OpenAI's Astra alignment claim, critic says GPT-5.5 safer — GarrisonLovely · 2026-09-04
- Miles Brundage: Astra demos are crazy, Anthropic surely not far behind — Miles_Brundage · 2026-09-04
- Sean Taylor: 'Fast progress on eradicating hallucinations,' backed by realistic Astra eval — DavideCrapis · 2026-09-04
- Microsoft launches MAI-Transcribe-2, claiming 10x speed of GPT-Transcribe — ZacharyHuang12 · 2026-09-04
- Fable 5.1 likely matches Astra on CoT controllability, observers say — Miles_Brundage · 2026-09-04
- Researcher doubts Gemini outage reports: Google's in-house infra makes shared failure unlikely — generativist · 2026-09-04