Claude Sonnet 5 burns 71% context window on first image query
IDisplayAgility · reddit · 2026-08-30
A user reported that inserting a screenshot into a long-standing Claude Sonnet 5 session (3+ months of context) and asking a simple question consumed 71% of the session's context limit. Subsequent questions only used 1-2%. This suggests a massive initial computational overhead when processing new images or reloading long contexts.
More from Models
- OpenAI stealth-testing mysterious internal model 'Aeon' in Codex desktop app and Dots — cedric_chee · 2026-10-16
- Argon edges out astra as daily driver, but Claude still wins heavy coding work — iruletheworldmo · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- RoboVista benchmark: Gemini-4 Argon hits 81.9% on robotics understanding — berkeley_ai · 2026-10-16
- Dev reports decision model jev beats gpt-6-luna on quality in real training pipeline — genmon · 2026-10-16
- Google won't publish Argon 4 limits in AI Pro—users compare experiences — Consistent-Text-2662 · 2026-10-16