Overview
You may occasionally run into one of the issues below. Here’s how to resolve each one.Errors in chat
Please try again: thinking or redacted_thinking
An error that looks like this:
Please try again messages.9.content.25:thinkingorredacted_thinkingblocks in the latest assistant message cannot be modified. These blocks must remain as they were in the original response.

claude-4.5-sonnet or claude-4.5-opus. Creating a new chat resolves it.

Please try again: Invalid schema for function bash
An error that looks like this:
Please try again Something went wrong. Attempting recovery… Invalid schema for function ‘bash’: In context=(), ‘required’ is required to be supplied and to be an array including every key in properties. Missing ‘timeout’.

gpt-5 or other OpenAI models (e.g.: o3, o3-pro, o4, etc). Anthropic models are unaffected (e.g.: claude-4.5-sonnet, claude-4.5-opus, etc).
To resolve:
- Switch models. Try switching from
gpt-5to another model (for example,claude-4.5-sonnet). - If the error continues, create a new chat. Use the “New Chat” button ( icon, top-left in chat panel) to start fresh.
Please try again: prompt is too long
An error that looks like this:Please try again prompt is too long: 207371 tokens > 200000 maximum

claude-4.5-sonnet or claude-4.5-opus.
To resolve: Create a new chat.

Endless generation
Sometimes Same may keep generating without finishing (the response never settles). To resolve:- Start a new chat using the “New Chat” button ( icon, top-left in the chat panel).
- Before you send your message, remove attachments you do not need. You can keep the chat reference attachment so Same still has context from the earlier conversation.

- If endless generation continues, remove every attachment (including the chat reference) and try again.
For troubleshooting prompts and debugging issues with your website, see Troubleshooting.