Junie provides error code 400: input length and `max_tokens` exceed context limit

Problem

A query to Junie fails with the error:

LLM returned error response:Exception during request to https://api.jetbrains.ai/user/v5/llm/chat/stream/v7 with code 400: Exception during request to https://all.agg.lim.internal.eu-west-1.prod.aws.ai.intellij.net/service/v5/chat/stream/v7 with code 400: input length and `max_tokens` exceed context limit: 197174 + 8192 > 200000, decrease input length or `max_tokens` and try again

Cause

The error may appear when too much context is sent to an LLM.

Resolution

Until a permanent fix is released, always start a new task by default.

Initiating follow-ups - especially in large numbers - can significantly increase the context size, increased latency, and a greater likelihood of hallucinations or inaccuracies. Only use follow-ups when it is absolutely necessary to reference specific details from a previous task. For further tips and guidance, visit When to use Junie follow-up prompt.