Problemsby IbisLlamaIndex runs out of context on real work
LLM apps2 posts from 1 person2d active+11 post in the last 7 days, 0 the 7 days before (steady)Unverified
Posts per day
The posts behind it
2, newest first[Bug]: SentenceSplitter can emit chunks over chunk_size after whitespace stripping### Bug Description `SentenceSplitter` can return chunks longer than `chunk_size`. This is the `SentenceSplitter` sibling of #22858, which covers `TokenTextSplitter`. `_merge()` budgets each chunk from the sum of the per-split token counts, but `_postprocess_chunks()` strips the...run-llama/llama_index · Harsh23KashyapOpen ↗[Bug]: Refine response synthesizer refines repacked sub-chunks in reverse order### Bug Description When `Refine` processes a chunk that no longer fits the prompt once the existing answer is included, the loop repacks the chunk into smaller sub-chunks and pushes them back to the front of a deque: `deque.extendleft` reverses the order of its argument. The...run-llama/llama_index · Harsh23Kashyap1 repliesOpen ↗
Companies and products named
About this problem
Evidence
2 posts from 1 person in 1 place, about 1 a week over 28 days. Mostly on run-llama/llama_index.
Frustration Frustration 0 of 3· Seen on GitHub issues
How it was grouped
Posts that state a pain and match the "context-window" rule about LlamaIndex. First post Sep 21, 2026, latest Oct 18, 2026. Unverified: fewer than 3 posts, or one person or place only. Method
The same problem elsewhere
Context windows run out on real workloadsClaude Code runs out of context on real workClaude runs out of context on real workChatGPT runs out of context on real workCursor runs out of context on real workOllama runs out of context on real workllama.cpp runs out of context on real workvLLM runs out of context on real workCodex runs out of context on real workQwen runs out of context on real workAnthropic runs out of context on real workOpenAI runs out of context on real work
Other LlamaIndex problems
LlamaIndex users keep filing bugsLlamaIndex upgrades break working setupsLlamaIndex fails to return valid structured outputTrouble with LlamaIndex returnsTrouble with LlamaIndex documentsTrouble with LlamaIndex eventsExperiments on LlamaIndex are hard to reproduce and trackTrouble with LlamaIndex raisesLlamaIndex integrations break or are missingRAG on LlamaIndex returns irrelevant or missing contextLlamaIndex token costs are hard to predictPrompt injection makes LlamaIndex features unsafe
History
LlamaIndex runs out of context on real work (2 posts)