What does "context length exceeded" or "message too long" mean?

It means the text you sent, plus any reply space reserved for the answer, is larger than the model can handle in one request. You must send less, split the work or summarize.

Fixes:

1. Split a long document into sections and process each separately.
2. Summarize earlier chat turns and start fresh with the summary.
3. Remove attachments or pasted text you do not need.
4. In the API, lower the maximum output setting. Input plus output has to fit together.
5. Count tokens before sending with a tokenizer tool from the vendor.
6. For big document sets, use retrieval (RAG) so only relevant passages are sent.

Why it happens: every model has a context window measured in tokens, and the window covers your prompt, history, attachments and the answer. Window sizes differ by model and change often, so check the vendor's current model documentation instead of relying on numbers you read elsewhere.