Claude Opus 4.6 Troubleshooting Checklist: Empty Replies, Upload Failures, and Quota Warnings

When using Claude Opus 4.6, if you run into issues like “can’t send,” “no reply,” or “can’t open files,” in most cases it’s not that the model is broken—it’s that your browser, network, or context length has hit a limit. Below is a symptom-based troubleshooting guide that can help you pinpoint the cause and restore the conversation within minutes.

Start with three basic troubleshooting steps—don’t rush to refresh repeatedly

Step 1: Check the network. Switch to a mobile hotspot or another connection and try again—this is the most time-saving troubleshooting step. Step 2: Switch browsers or use an incognito window. Many cases of “clicking the button does nothing” come from extension scripts or cache conflicts. Step 3: Clear the site cookies and log in again. An abnormal login state can make a request look like it was sent successfully while actually being blocked.

If you’re on a company/campus network, proxies, gateways, or content filtering can also cause intermittent failures. The troubleshooting criterion for this type is: same account, same issue—switch networks and it immediately works.

Message send failure / empty replies: troubleshoot in the order “content → conversation → page”

If it keeps spinning after you send or the reply comes back empty, first copy your input before troubleshooting: remove overly long quotes, large chunks of text pasted repeatedly, and unnecessary code blocks. Then start a new chat and retry—when an old conversation’s context gets too long, it’s more likely to get stuck during generation.

At the page level, troubleshoot again: disable ad blockers/script blockers, or temporarily turn off “auto-translate” type plugins. They can rewrite the page DOM and cause input box issues. Finally, refresh the page, and avoid clicking send repeatedly while generation is in progress—rapid consecutive sends can trigger short-term request congestion.

File upload failure / parse failure: troubleshoot starting from format and size

For file-related issues, it’s recommended to start with a “minimal test” troubleshooting step: use a small file in a common format (e.g., a clean PDF or TXT) to verify whether the upload pipeline is working. If small files work but large ones don’t, it’s usually because the file is too large, has too many pages, or contains scanned images that increase parsing load.

Do another round of troubleshooting: unzip archives first, merge multiple files into one, convert images to a more common format, and remove special characters from the filename. If it says parsing failed, try exporting a version where the text is selectable. For scanned documents, run OCR before uploading—the success rate will improve significantly.

Quota/permission warnings and model unavailable: troubleshooting usage and account status

If you see a quota warning or “currently unavailable,” first troubleshoot whether it’s peak-time congestion: switch to a different chat, shorten the length of a single request, and break “write a long article” into an outline and generate it in sections. If you keep hitting limits, check whether you’re using multiple devices at the same time or retrying frequently—high-density requests within the same time window are more likely to be throttled by the system.

If you see permission-related warnings, the troubleshooting focus is account status: log out and back in, and confirm the browser hasn’t auto-switched you to an old account. If it still doesn’t recover, keep a screenshot of the error and the approximate time, so you can quickly pinpoint the issue later when you self-check or submit feedback.