Brief
LlamaIndex v0.14.14 adds a token-budget callback and LangChain 1.x support
The v0.14.14 release notes for LlamaIndex list llama-index-core at 0.14.14 with a new TokenBudgetHandler callback filed under cost governance, plus LangChain 1.x support and a set of core fixes.
The release notes for v0.14.14 record llama-index-core at 0.14.14 and include the changelog entry "feat(callbacks): add TokenBudgetHandler for cost governance (#20546)". The same notes list "Langchain1.x support (#20472)", a fix catching pydantic ValidationError in VectorStoreQueryOutputParser (#20450), a fix distinguishing empty string from None in MediaResource.hash (#20451), a retry-logic fix for empty LLM responses in agents (#20596), and, in core components, "Fix potential crashes and improve security defaults" (#20610).
Our reading
For readers who already run LlamaIndex pipelines with callbacks attached, a named token-budget handler points at spend control, and cost is usually the item teams try to add after an agent is already making metered calls. The LangChain 1.x line and the security-defaults fix in core components are the two other entries worth noting, because both touch code a reader may already be maintaining. The…
What to do or watch
Before adopting, resolve what TokenBudgetHandler actually does: whether it enforces a limit or only counts tokens, what unit the budget is set in, and what happens when it is exceeded. The release note does not state this.
Source details and supporting facts
Each line is stated by the page named above it.
Stated by github.com
- llama-index-core is version 0.14.14 in this release.
- The changelog entry "feat(callbacks): add TokenBudgetHandler for cost governance ( #20546 )" appears in the release notes.
- The release notes include "Langchain1.x support ( #20472 )".
- The release notes include "Fix potential crashes and improve security defaults in core components ( #20610 )".
- The release notes include "fix: catch pydantic ValidationError in VectorStoreQueryOutputParser ( #20450 )".
- The release notes include "fix(agent): handle empty LLM responses with retry logic and add test cases ( #20596 )".
Sources
- LlamaIndex releasesText stored 17 September 2026
How this story was checked. Written from the 1 page listed above, stored 17 September 2026; claims checked against that stored text on 17 September 2026.
What that means
- 6 of 6 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.