Chat with content: hitting TPM limits very fast, even with quite basic static content (R markdown) - can chunking be used

Dear,

I have deployed "chat with content" and I am trying to use it on (quite minimal) R markdown reports. I have set "chat with content" up to use OpenAI - I tried multiple iterations with gpt-4.1-mini, gpt-5-mini, ... but I am always receiving an error: Request too large for gpt-4.1-long-context in organization <my-org> on tokens per min (TPM): Limit 500000, Requested <number larger than 500k>. The input or output tokens must be reduced in order to run successfully. Visit https://platform.openai.com/account/rate-limits to learn more.', 'type': 'tokens', 'param': None, 'code': 'rate_limit_exceeded'}}.

I am a bit surprised about the number of tokens that even relatively basic R markdown reports take up, but I am wondering:

• Why, regardless of what I specify under CHATLAS_CHAT_ARGS, do I always see gpt-4.1-long-context in the error message?
• Is there a way to implement some kind of RAG "chunking" to avoid exceeding rate limits?

Thanks in advence,

FM Kerckhof

Start by tracing how CHATLAS_CHAT_ARGS selects the OpenAI model in the chat-with-content extension, then reproduce the TPM error with a minimal R Markdown report. Check the existing content retrieval path for any chunking behavior. Done should include a clear explanation of the model selection and a defined approach for handling reports that exceed TPM limits.