Brief
llama-index v0.14.21 patches node deletion, structured-output errors and persistence encoding in core
The llama-index v0.14.21 release notes list llama-index-core [0.14.21] with several fixes and one integration addition, but carry no migration notes or test details, so the practical impact is inferred from the change descriptions alone.
The release notes for llama-index v0.14.21 list llama-index-core [0.14.21] alongside a spread of integration packages. Among them, llama-index-embeddings-nvidia [0.5.1] adds HTTP client support, and a long series of dependency bumps touches the pip group across dozens of directories.
The core changes are maintenance-shaped: a fix to prevent a KeyError in DocumentSummaryIndex.delete_nodes when an invalid node ID is provided, handling for ValueError and TypeError raised by structured output failures, explicit UTF-8 encoding added to persistence layer fs.open() calls, and a fix for a breaking change in message block buffer resolution.
The material does not state migration steps, test instructions or reproduction cases for any of these changes.
Our reading
Two of these touch steps that quietly derail pipelines: deleting nodes from a summary index and handling malformed structured output. Teams running document-summary indexes or structured extraction should treat both as candidates for a targeted regression test, while the UTF-8 persistence change matters most where workflows write files across platforms. The NVIDIA embedding package gaining HTTP c…
What to do or watch
Run an invalid node ID through delete_nodes and a deliberately malformed structured-output response against a canary install before promoting; the notes do not say whether either condition previously crashed the run or was silently skipped.
Source details and supporting facts
Each line is stated by the page named above it.
Stated by github.com
- llama-index-core [0.14.21] fixes a KeyError in DocumentSummaryIndex.delete_nodes when an invalid node ID is provided (#21067).
- llama-index-core [0.14.21] handles ValueError and TypeError from structured output failures (#21090).
- llama-index-core [0.14.21] adds explicit UTF-8 encoding to persistence layer fs.open() calls (#21111).
- llama-index-core [0.14.21] includes a fix titled "Fix Breaking Change in Message Block Buffer Resolution" (#21339).
- llama-index-embeddings-nvidia [0.5.1] adds HTTP client support (#21046).
Sources
- LlamaIndex releasesText stored 16 September 2026
How this story was checked. Written from the 1 page listed above, stored 16 September 2026; claims checked against that stored text on 17 September 2026.
What that means
- 5 of 5 reported statements were confirmed against the page that carries them; the rest were removed rather than published.
- Figures in the text were required to appear in the stored source text: yes. Identifiers: yes.
- The check reads stored text only: no claim rests on a fresh look that did not happen.
- Where the reporting was silent, the text says so instead of filling the gap.