LlamaIndex token costs are hard to predict

Cloud and tool cost2 posts from 2 people2d active+00 posts in the last 7 days, 1 the 7 days before (fading)Unverified

Posts per day2 posts, Sep 7 to Oct 21

The posts behind it

2, newest first

GitHub issuesOct 14 [Bug]: AG-UI router: system_prompt is emitted in MESSAGES_SNAPSHOT and grows by one copy every run### Bug Description With `get_ag_ui_workflow_router(..., system_prompt=...)`, the server's system prompt ends up in the client's thread and is added again on every run, so the prompt the LLM receives grows by one copy per run. `AGUIChatWorkflow.chat` inserts...run-llama/llama_index · LaptopminiOpen ↗GitHub issuesSep 7 [Feature Request]: Idea: token-budgeted connected subgraph retrieval### Feature Description Hi! First of all, thanks for building LlamaIndex! I maintain [SteinerPy](https://github.com/berendmarkhorst/SteinerPy), a Python package for Steiner tree problems and related graph optimization problems. While looking at PropertyGraphIndex, I wondered...run-llama/llama_index · berendmarkhorst1 repliesOpen ↗

Companies and products named

Named2 posts

LlamaIndex Named1 post

GitHub

About this problem

Evidence

2 posts from 2 people in 1 place, about 1 a week over 38 days. Mostly on run-llama/llama_index. Tools named alongside: GitHub.

Frustration Frustration 0 of 3· Seen on GitHub issues

How it was grouped

Posts that state a pain and match the "llm-cost" rule about LlamaIndex. First post Sep 7, 2026, latest Oct 14, 2026. Unverified: fewer than 3 posts, or one person or place only. Method

The same problem elsewhere

Elsewhere11 postsLLM token costs are hard to predict and control Elsewhere4 postsAnthropic token costs are hard to predict Elsewhere3 postsClaude token costs are hard to predict Elsewhere2 postsGemini token costs are hard to predict

Other LlamaIndex problems

Same product15 postsLlamaIndex users keep filing bugs Same product10 postsLlamaIndex upgrades break working setups Same product5 postsLlamaIndex fails to return valid structured output Same product3 postsTrouble with LlamaIndex returns Same product3 postsTrouble with LlamaIndex documents Same product3 postsTrouble with LlamaIndex events Same product3 postsExperiments on LlamaIndex are hard to reproduce and track Same product2 postsLlamaIndex runs out of context on real work Same product2 postsTrouble with LlamaIndex raises Same product2 postsLlamaIndex integrations break or are missing Same product2 postsRAG on LlamaIndex returns irrelevant or missing context Same product2 postsPrompt injection makes LlamaIndex features unsafe

History

MomentumOct 21, 2026steady to fading

AddedOct 19, 2026LlamaIndex token costs are hard to predict (2 posts)