Too much data to handle - OpenAIModelTokenLimit | Copilot Studio and SharePoint

Question

Too much data to handle - OpenAIModelTokenLimit | Copilot Studio and SharePoint

Marco Riezzo 45

After setting a Copilot Studio Agent with a SharePoint knowledge base containing 12 files (from 1MB to 32MB, with multiple MS 365 licenses so widely under the 200MB limit in place in this case), ANY kind of question (regardless how much generic it is), results in the following error (TooMuchDataToHandle):

Error Message: The request is resulting in too much data to handle, please evaluate the amount of data being returned by your actions.[Learn more]https://learn.microsoft.com/en-us/microsoft-copilot-studio/error-codes?tabs=webApp#toomuchdatatohandleError Code: OpenAIModelTokenLimit Conversation Id: <id removed> Time (UTC): 2026-02-03T13:51:37.154Z

This is something that started to happen in the last 1-2 days and before that time, all the files (the same files described above) were indexed and retrieved successfully in SharePoint.

Jiri Sedivec 30 Punti di reputazione

2026-02-04T08:54:28.21+00:00
Hi everyone,

We’re seeing the same issue today in Copilot Studio: “Too much data to handle” / OpenAIModelTokenLimit. This started occurring only today — we didn’t have it previously with the same knowledge sources and the same prompts.

Scenario (repro):

Agent in Copilot Studio using GPT-4.1

Knowledge sources: SharePoint-based knowledge (plus our existing indexed content)

Simple user question like: “What categories do we use in Documents?”

The agent fails immediately with the error, even in a fresh test session.

We tried small mitigations (e.g., limiting how many documents we list in the final answer), but it looks like the failure happens before response generation (likely during retrieval/grounding) because the request becomes too large.

Has anyone found a workaround or heard of a service-side regression/incident affecting knowledge retrieval payload size today?

Thanks!
Duong Pham 5 Punti di reputazione

2026-02-23T06:47:44.67+00:00

I'm facing the same issue. While the agent created in M365 Copilot Chat is working well, the agents created in Copilot Studio with license get stuck with this problem.

1 risposta

Risposta

Jiri Sedivec 30 Punti di reputazione

2026-02-04T08:54:28.21+00:00

Hi everyone,

We’re seeing the same issue today in Copilot Studio: “Too much data to handle” / OpenAIModelTokenLimit. This started occurring only today — we didn’t have it previously with the same knowledge sources and the same prompts.

Scenario (repro):

Agent in Copilot Studio using GPT-4.1

Knowledge sources: SharePoint-based knowledge (plus our existing indexed content)

Simple user question like: “What categories do we use in Documents?”

The agent fails immediately with the error, even in a fresh test session.

We tried small mitigations (e.g., limiting how many documents we list in the final answer), but it looks like the failure happens before response generation (likely during retrieval/grounding) because the request becomes too large.

Has anyone found a workaround or heard of a service-side regression/incident affecting knowledge retrieval payload size today?

Thanks!
Duong Pham 5 Punti di reputazione

2026-02-23T06:47:44.67+00:00

I'm facing the same issue. While the agent created in M365 Copilot Chat is working well, the agents created in Copilot Studio with license get stuck with this problem.

Answer 1

Jim Huguelet 5

We are seeing the same thing ("OpenAIModelTokenLimit") although in our case many generative answers are still being created - but we had never seen this situation before today. I tried switching from GTP-4.1 to the four available GTP-5 models and all produce the behavior with roughly the same frequency.

Pablo Arisi 10 Punti di reputazione

2026-02-06T14:10:05.3233333+00:00

exactly the same problem, today..

Condividi tramite

Too much data to handle - OpenAIModelTokenLimit | Copilot Studio and SharePoint

1 risposta

Risposta