This release adds the newest ChatGPT models and makes long browsing runs much lighter on your subscription's usage limit.
Long runs use less of your limit
Some of you found that a single long run, like collecting hundreds of listings, used up your whole ChatGPT allowance. Every step of a run resends the conversation so far, so we've cut how much each step carries:
- Earlier, cheaper compaction — Long chats are summarized sooner, and the summary is now read from the provider's cache instead of being sent again in full.
- Fewer old screenshots — Each step only carries your most recent screenshots. Older ones stay in your saved chat, and the agent takes a new one when it needs to look again.
- Smaller page snapshots and screenshots — Tracking parameters are stripped from links, long tool output is saved to a file instead of kept in the chat, and screenshots are sent at a smaller size.
- Medium thinking by default — New installs start at medium thinking, which is plenty for step-by-step browsing. You can still pick high in the message box.
ChatGPT models
- GPT-6.1 Terra and GPT-6.1 Sol — Now available with your ChatGPT subscription. GPT-6.1 Terra is the new default.
- Model moves — If you had GPT-5.6 Sol or GPT-6 Sol selected, you move to GPT-6.1 Sol; GPT-5.6 Terra, GPT-5.5 and GPT-5.4 move to GPT-6.1 Terra. If you're on GPT-6 Astra, consider GPT-6.1 Sol or Terra for long runs: Astra uses your limit fastest.
Questions or feedback? Reach out at sawyer@dobrowser.io.