# LlamaIndex changelog

> A data framework for connecting custom data sources to large language models.

- Vendor: LlamaIndex
- Category: AI
- Official site: https://www.llamaindex.ai
- Tracked by: What's New (https://whatsnew.fyi/product/llamaindex)
- Harvested from: GitHub (run-llama/llama_index)
- Entries below: 10 (newest first)

What's New is an index, not a publisher: every entry below links to the vendor's own release notes, which are the authoritative source. Entries are labelled where they are hand-curated sample data, pre-releases, or drawn from a secondary source such as a developer blog.

Reuse: the summaries, labels and curation here are © What's New. Quote freely with attribution and a link back; wholesale republication of the corpus is not permitted — terms: https://whatsnew.fyi/terms. The vendors' own release notes remain their publishers'.

## Releases

### v0.14.23

- Date: 2026-06-24
- Version: v0.14.23
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.23
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.23

- **added** — Multimodal synthesis capabilities
- **added** — Tool calling mock LLM
- **added** — Multimodal query engines
- **added** — DocumentBlock and VideoBlock support in FunctionTool output parsing
- **fixed** — Guard against ZeroDivisionError on empty input sequences in prompt_helper
- **fixed** — Preserve URL-backed video and document memory blocks
- **fixed** — Add explicit encoding='utf-8' to llama-index-core text-mode file I/O
- **fixed** — Preserve IndexNode obj during model dump
- **fixed** — Deep copy initial_state in workflow to prevent mutation leaks across runs
- **fixed** — Match missing metadata for NE and NIN filters
- **fixed** — Preserve TreeSelectLeafRetriever source nodes
- **fixed** — Fix RecursionError in TokenTextSplitter and SentenceSplitter for units larger than chunk_size
- **fixed** — Fix refresh_ref_docs kwargs reuse
- **fixed** — Use running loop in ingestion pipeline
- **changed** — Use a set instead of a list for within-batch dedup in ingestion pipeline
- **changed** — Simplify serialized payloads to instrumentation
- **changed** — Allow google-genai 2.0+ SDKs
- **fixed** — Use utf-8 for adapter config files in llama-index-embeddings-adapter

#### Release Notes

##### [2026-06-24]

###### llama-index-callbacks-argilla [0.5.0]

- chore(deps): bump the uv group across 32 directories with 3 updates ([#21664](https://github.com/run-llama/llama_index/pull/21664))
- chore(deps): bump the uv group across 29 directories with 3 updates ([#21665](https://github.com/run-llama/llama_index/pull/21665))
- chore(deps): bump the uv group across 31 directories with 3 updates ([#21668](https://github.com/run-llama/llama_index/pull/21668))
- chore(deps): bump the pip group across 35 directories with 4 updates ([#21714](https://github.com/run-llama/llama_index/pull/21714))
- chore(deps): bump the uv group across 32 directories with 5 updates ([#21717](https://github.com/run-llama/llama_index/pull/21717))
- chore(deps): bump the uv group across 31 directories with 5 updates ([#21719](https://github.com/run-llama/llama_index/pull/21719))
- chore(deps): bump the uv group across 31 directories with 5 updates ([#21720](https://github.com/run-llama/llama_index/pull/21720))
- chore(deps): bump the uv group across 30 directories with 5 updates ([#21721](https://github.com/run-llama/llama_index/pull/21721))
- chore(deps): bump the uv group across 30 directories with 5 updates ([#21722](https://github.com/run-llama/llama_index/pull/21722))
- chore(deps): bump the pip group across 12 directories with 4 updates ([#21724](https://github.com/run-llama/llama_index/pull/21724))
- chore(deps): bump the uv group across 21 directories with 3 updates ([#21725](https://github.com/run-llama/llama_index/pull/21725))
- chore(deps): bump the pip group across 11 directories with 4 updates ([#21726](https://github.com/run-llama/llama_index/pull/21726))
- chore(deps): bump the uv group across 17 directories with 3 updates ([#21727](https://github.com/run-llama/llama_index/pull/21727))
- chore(deps): bump the uv group across 18 directories with 4 updates ([#21733](https://github.com/run-llama/llama_index/pull/21733))
- chore(deps): bump the uv group across 7 directories with 6 updates ([#21736](https://github.com/run-llama/llama_index/pull/21736))

###### llama-index-core [0.14.23]

- feat(core): Multimodal synthesis part 2 ([#21561](https://github.com/run-llama/llama_index/pull/21561))
- fix(core): add DocumentBlock and VideoBlock to FunctionTool.\_parse_tool_output ([#21678](https://github.com/run-llama/llama_index/pull/21678))
- fix(prompt_helper): guard against ZeroDivisionError on empty input sequences ([#21707](https://github.com/run-llama/llama_index/pull/21707))
- Preserve URL-backed video and document memory blocks ([#21728](https://github.com/run-llama/llama_index/pull/21728))
- fix: add explicit encoding='utf-8' to llama-index-core text-mode file I/O ([#21729](https://github.com/run-llama/llama_index/pull/21729))
- Add tool calling mock LLM ([#21732](https://github.com/run-llama/llama_index/pull/21732))
- make tests green again ([#21737](https://github.com/run-llama/llama_index/pull/21737))
- Fix refresh_ref_docs kwargs reuse ([#21752](https://github.com/run-llama/llama_index/pull/21752))
- perf: use a set instead of a list for within-batch dedup in Ingestion… ([#21755](https://github.com/run-llama/llama_index/pull/21755))
- fix: use running loop in ingestion pipeline ([#21765](https://github.com/run-llama/llama_index/pull/21765))
- fix(core): preserve IndexNode obj during model dump ([#21776](https://github.com/run-llama/llama_index/pull/21776))
- fix(workflow): deep copy initial_state to prevent mutation leaks across runs ([#21780](https://github.com/run-llama/llama_index/pull/21780))
- Multimodal query engines ([#21784](https://github.com/run-llama/llama_index/pull/21784))
- fix(core): match missing metadata for NE and NIN filters ([#21785](https://github.com/run-llama/llama_index/pull/21785))
- fix: preserve TreeSelectLeafRetriever source nodes ([#21787](https://github.com/run-llama/llama_index/pull/21787))
- Fix RecursionError in TokenTextSplitter & SentenceSplitter for units larger than chunk_siz

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.23]_

### v0.14.22

- Date: 2026-05-14
- Version: v0.14.22
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.22
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.22

- **fixed** — Let SparseEmbeddingStartEvent inherit EmbeddingStartEvent
- **fixed** — Remove stale exports from public __all__ lists
- **fixed** — Preserve cache writes from multiprocessing workers in IngestionPipeline
- **added** — Add multimodal synthesis support
- **fixed** — Avoid mutating response.raw in LLM event model_dump methods
- **fixed** — Handle DocumentBlock in Memory._estimate_token_count
- **fixed** — Propagate contextvars in sync_to_async for FunctionTool

#### Release Notes

##### [2026-05-14]

###### llama-index-agent-agentmesh [0.2.0]

- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-agentops [0.5.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-aim [0.4.1]

- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-argilla [0.5.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))

###### llama-index-callbacks-arize-phoenix [0.7.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-honeyhive [0.5.0]

- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-langfuse [0.5.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-literalai [1.4.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-openinference [0.5.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-opik [1.3.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-promptlayer [0.5.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-uptrain [0.6.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-callbacks-wandb [0.5.0]

- chore(deps): bump the pip group across 55 directories with 3 updates ([#21435](https://github.com/run-llama/llama_index/pull/21435))
- mass uv lock --upgrade ([#21638](https://github.com/run-llama/llama_index/pull/21638))

###### llama-index-core [0.14.22]

- fix(instrumentation): let SparseEmbeddingStartEvent inherit EmbeddingStartEvent ([#21119](https://github.com/run-llama/llama_index/pull/21119))
- fix: remove stale exports from public **all** lists ([#21133](https://github.com/run-llama/llama_index/pull/21133))
- Preserve cache writes from multiprocessing workers in IngestionPipeline ([#21301](https://github.com/run-llama/llama_index/pull/21301))
- feat(core): Multimodal synthesis ([#21374](https://github.com/run-llama/llama_index/pull/21374))
- fix: avoid mutating response.raw in LLM event model_dump methods ([#21424](https://github.com/run-llama/llama_index/pull/21424))
- fix(memory): handle DocumentBlock in Memory.\_estimate_token_count ([#21529](https://github.com/run-llama/llama_index/pull/21529))
- fix: propagate contextvars in sync_to_async for FunctionTool ([#21558](https://github.com/run-llama/llama_index/pull/2

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.22]_

### v0.14.21

- Date: 2026-04-21
- Version: v0.14.21
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.21
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.21

- **fixed** — prevent KeyError in DocumentSummaryIndex.delete_nodes when invalid node ID is provided
- **fixed** — handle ValueError and TypeError from structured output failures
- **fixed** — add explicit UTF-8 encoding to persistence layer fs.open() calls
- **fixed** — fix breaking change in message block buffer resolution
- **added** — add http client support to embeddings-nvidia

#### Release Notes

##### [2026-04-21]

###### llama-index-callbacks-honeyhive [0.5.0]

- chore(deps): bump the pip group across 87 directories with 2 updates ([#21382](https://github.com/run-llama/llama_index/pull/21382))
- chore(deps): bump the pip group across 68 directories with 2 updates ([#21394](https://github.com/run-llama/llama_index/pull/21394))

###### llama-index-core [0.14.21]

- fix(core): prevent `KeyError` in `DocumentSummaryIndex.delete_nodes` when invalid node ID is provided ([#21067](https://github.com/run-llama/llama_index/pull/21067))
- fix(core): handle `ValueError` and `TypeError` from structured output failures ([#21090](https://github.com/run-llama/llama_index/pull/21090))
- fix: add explicit UTF-8 encoding to persistence layer fs.open() calls ([#21111](https://github.com/run-llama/llama_index/pull/21111))
- Fix Breaking Change in Message Block Buffer Resolution ([#21339](https://github.com/run-llama/llama_index/pull/21339))
- chore(deps): bump the pip group across 87 directories with 2 updates ([#21382](https://github.com/run-llama/llama_index/pull/21382))
- chore(deps): bump the pip group across 68 directories with 2 updates ([#21394](https://github.com/run-llama/llama_index/pull/21394))

###### llama-index-embeddings-huggingface-optimum [0.4.1]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-nvidia [0.5.1]

- feat(embeddings-nvidia): add http client support ([#21046](https://github.com/run-llama/llama_index/pull/21046))

###### llama-index-embeddings-oracleai [0.4.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-premai [0.5.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))
- chore(deps): bump the pip group across 87 directories with 2 updates ([#21382](https://github.com/run-llama/llama_index/pull/21382))
- chore(deps): bump the pip group across 68 directories with 2 updates ([#21394](https://github.com/run-llama/llama_index/pull/21394))

###### llama-index-embeddings-text-embeddings-inference [0.5.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-textembed [0.4.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-together [0.5.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-upstage [0.6.1]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))
- chore(deps): bump the pip group across 87 directories with 2 updates ([#21382](https://github.com/run-llama/llama_index/pull/21382))
- chore(deps): bump the pip group across 68 directories with 2 updates ([#21394](https://github.com/run-llama/llama_index/pull/21394))

###### llama-index-embeddings-vertex [0.5.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-vertex-endpoint [0.4.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-vllm [0.1.1]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama-index-embeddings-voyageai [0.6.0]

- chore(deps): bump the pip group across 96 directories with 2 updates ([#21381](https://github.com/run-llama/llama_index/pull/21381))

###### llama

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.21]_

### v0.14.20

- Date: 2026-04-03
- Version: v0.14.20
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.20
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.20

- **security** — Fix vulnerability with nltk across multiple packages

#### Release Notes

##### [2026-04-03]

###### llama-index-agent-agentmesh [0.2.0]

- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-agentops [0.5.0]

- chore(deps): bump the uv group across 50 directories with 2 updates ([#21164](https://github.com/run-llama/llama_index/pull/21164))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llama/llama_index/pull/21221))
- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-aim [0.4.1]

- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-argilla [0.5.0]

- chore(deps): bump the uv group across 58 directories with 1 update ([#21166](https://github.com/run-llama/llama_index/pull/21166))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llama/llama_index/pull/21221))
- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-arize-phoenix [0.7.0]

- chore(deps): bump the uv group across 58 directories with 1 update ([#21166](https://github.com/run-llama/llama_index/pull/21166))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llama/llama_index/pull/21221))
- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-honeyhive [0.5.0]

- chore(deps): bump the uv group across 58 directories with 1 update ([#21166](https://github.com/run-llama/llama_index/pull/21166))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llama/llama_index/pull/21221))
- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-langfuse [0.5.0]

- chore(deps): bump the uv group across 58 directories with 1 update ([#21166](https://github.com/run-llama/llama_index/pull/21166))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llama/llama_index/pull/21221))
- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-literalai [1.4.0]

- chore(deps): bump the uv group across 58 directories with 1 update ([#21166](https://github.com/run-llama/llama_index/pull/21166))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llama/llama_index/pull/21221))
- fix vulnerability with nltk ([#21275](https://github.com/run-llama/llama_index/pull/21275))

###### llama-index-callbacks-openinference [0.5.0]

- chore(deps): bump the uv group across 58 directories with 1 update ([#21166](https://github.com/run-llama/llama_index/pull/21166))
- chore(deps): bump the uv group across 24 directories with 1 update ([#21219](https://github.com/run-llama/llama_index/pull/21219))
- chore(deps): bump the uv group across 21 directories with 2 updates ([#21221](https://github.com/run-llam

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.20]_

### v0.14.19

- Date: 2026-03-25
- Version: v0.14.19
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.19
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.19

- **fixed** — Pass delete_from_docstore parameter in BaseIndex.delete_ref_doc
- **fixed** — Preserve CTE names during schema prefixing in SQLDatabase.run_sql
- **fixed** — Align sync retrieval dedup key with async using hash + ref_doc_id
- **fixed** — Raise ValueError instead of returning string from structured_predict
- **fixed** — Remove incorrect per-node delete calls in index helpers
- **fixed** — Fix llama-cloud managed index and remove llamaparse reader
- **added** — Support azure openai responses
- **fixed** — Fix azure openai responses
- **fixed** — Use proper tool choice format in bedrock converse
- **added** — Set gemini 3 as default and add temperature support in google-genai
- **fixed** — Avoid mutating messages list in prepare_chat_params for google-genai
- **added** — Add support for custom LLM provider in model kwargs for litellm
- **added** — Add MiniMax LLM provider integration with M2.7 default
- **fixed** — Pass custom headers to auto-created clients in ollama
- **added** — Add support for Mini and Nano variants of GPT 5.4 in openai
- **changed** — Enable llama-cloud>1.0 install

#### Release Notes

##### [2026-03-25]

###### llama-index-agent-agentmesh [0.2.0]

- chore(deps): bump the uv group across 49 directories with 1 update ([#21083](https://github.com/run-llama/llama_index/pull/21083))

###### llama-index-callbacks-argilla [0.5.0]

- chore(deps): bump the uv group across 3 directories with 1 update ([#21069](https://github.com/run-llama/llama_index/pull/21069))

###### llama-index-core [0.14.19]

- fix: pass `delete_from_docstore` parameter in `BaseIndex.delete_ref_doc` ([#20990](https://github.com/run-llama/llama_index/pull/20990))
- fix(core): preserve CTE names during schema prefixing in SQLDatabase.run_sql ([#21028](https://github.com/run-llama/llama_index/pull/21028))
- fix(core): align sync retrieval dedup key with async (hash + ref_doc_id) ([#21034](https://github.com/run-llama/llama_index/pull/21034))
- fix(core): raise ValueError instead of returning string from structured_predict ([#21036](https://github.com/run-llama/llama_index/pull/21036))
- fix(core): remove incorrect per-node delete calls in index helpers ([#21050](https://github.com/run-llama/llama_index/pull/21050))
- chore(deps): bump the uv group across 49 directories with 1 update ([#21083](https://github.com/run-llama/llama_index/pull/21083))
- chore(deps): bump the uv group across 44 directories with 1 update ([#21097](https://github.com/run-llama/llama_index/pull/21097))
- enable llama-cloud>1.0 install ([#21140](https://github.com/run-llama/llama_index/pull/21140))

###### llama-index-embeddings-fireworks [0.5.2]

- test(embeddings-fireworks): add test suite and fix docs ([#20977](https://github.com/run-llama/llama_index/pull/20977))

###### llama-index-embeddings-upstage [0.6.1]

- chore(deps): bump the uv group across 49 directories with 1 update ([#21083](https://github.com/run-llama/llama_index/pull/21083))

###### llama-index-indices-managed-llama-cloud [0.11.1]

- fix: llama-cloud managed index and remove llamaparse reader ([#21043](https://github.com/run-llama/llama_index/pull/21043))
- enable llama-cloud>1.0 install ([#21140](https://github.com/run-llama/llama_index/pull/21140))

###### llama-index-llms-azure-openai [0.5.3]

- azure openai responses support ([#21088](https://github.com/run-llama/llama_index/pull/21088))
- fix azure openai responses ([#21099](https://github.com/run-llama/llama_index/pull/21099))

###### llama-index-llms-bedrock-converse [0.14.3]

- use proper tool choice format in bedrock converse ([#21098](https://github.com/run-llama/llama_index/pull/21098))

###### llama-index-llms-cohere [0.8.0]

- docs(cohere): update first basic usage example to chat API ([#21108](https://github.com/run-llama/llama_index/pull/21108))

###### llama-index-llms-google-genai [0.9.1]

- feat: gemini 3 default and temperature ([#21060](https://github.com/run-llama/llama_index/pull/21060))
- fix(google-genai): avoid mutating messages list in prepare_chat_params ([#21141](https://github.com/run-llama/llama_index/pull/21141))

###### llama-index-llms-litellm [0.7.1]

- Add support for custom LLM provider in model kwargs ([#21095](https://github.com/run-llama/llama_index/pull/21095))

###### llama-index-llms-minimax [0.1.0]

- feat: add MiniMax LLM provider integration with M2.7 default ([#20955](https://github.com/run-llama/llama_index/pull/20955))

###### llama-index-llms-ollama [0.10.1]

- fix(ollama): pass custom headers to auto-created clients ([#21091](https://github.com/run-llama/llama_index/pull/21091))

###### llama-index-llms-openai [0.7.3]

- feat(llms/openai): Add support for Mini and Nano variants of GPT 5.4 ([#21065](https://github.com/run-llama/llama_index/pull/21065))

###### llama-index-llms-ovhcloud [0.2.1]

- chore(deps): bump the uv group across 49 directories with 1 update ([#21083](https://github.com/run-llama/llama_index/pull/21083))

###### llama-index-packs-agent-search-retriever [0.5.0]

- chore(deps): bump the uv group across 49 directories with 1 update ([#21083](https://github.com/run-llama

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.19]_

### v0.14.18

- Date: 2026-03-16
- Version: v0.14.18
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.18
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.18

- **added** — Align text match filters across core and vector backends
- **fixed** — Preserve chat history on incomplete stream consumption
- **fixed** — Guard against ZeroDivisionError in LlamaDebugHandler._get_time_stats_from_event_pairs
- **fixed** — Add stacklevel=2 to warnings.warn() for accurate caller reporting
- **deprecated** — Deprecate Python 3.9
- **fixed** — Use postprocess_nodes() in async retrieval paths
- **fixed** — Preserve response metadata in async _aretrieve_from_object
- **fixed** — Preserve non-ASCII schema descriptions in PydanticOutputParser
- **fixed** — structured_predict() returns default values for single-field models
- **fixed** — Fix OpenAI mimetype guess

#### Release Notes

##### [2026-03-16]

###### llama-index-agent-agentmesh [0.2.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-agent-azure [0.3.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-agentops [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-argilla [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-arize-phoenix [0.7.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-honeyhive [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))
- chore(deps): bump the uv group across 51 directories with 3 updates ([#20962](https://github.com/run-llama/llama_index/pull/20962))

###### llama-index-callbacks-langfuse [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-literalai [1.4.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-openinference [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-opik [1.3.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-promptlayer [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-uptrain [0.6.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-callbacks-wandb [0.5.0]

- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))

###### llama-index-core [0.14.18]

- feat: align text match filters across core and vector backends ([#20883](https://github.com/run-llama/llama_index/pull/20883))
- fix(chat_engine): preserve chat history on incomplete stream consumption ([#20897](https://github.com/run-llama/llama_index/pull/20897))
- fix: guard against ZeroDivisionError in LlamaDebugHandler.\_get_time_stats_from_event_pairs ([#20937](https://github.com/run-llama/llama_index/pull/20937))
- fix: add stacklevel=2 to warnings.warn() for accurate caller reporting ([#20939](https://github.com/run-llama/llama_index/pull/20939))
- chore: deprecate python 3.9 once and for all ([#20956](https://github.com/run-llama/llama_index/pull/20956))
- Release 0.14.17 ([#20957](https://github.com/run-llama/llama_index/pull/20957))
- chore(deps): bump the uv group across 51 directories with 3 updates ([#20962](https://github.com/run-llama/llama_index/pull/20962))
- fix: use `apostprocess_nodes()` in async retrieval paths ([#20974](https://github.com/run-llama/llama_index/pull/20974))
- fix (test) : use >= 1 to avoid racy `stream_chat` memory assertion ([#20980](https://github.com/run-llama/llama_index/pull/20980))
- fix(core): preserve response metadata in `async _aretrieve_from_object` ([#20995](https://github.com/run-llama/llama_index/pull/20995))
- fix: preserve non-ASCII schema descriptions in `PydanticOutputParser` ([#21016](https://github.com/run-llama/llama_index/pull/21016))
- chore(deps): bump the uv group across 42 directories with 2 updates ([#21020](https://github.com/run-llama/llama_index/pull/21020))
- fix(core): structured_predict() returns default values for single-field models ([#21025](https://github.com/run-llama/llama_index/pull/21025))
- fix openai mimetype guess (

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.18]_

### v0.14.16

- Date: 2026-03-10
- Version: v0.14.16
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.16
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.16

- **added** — Add token-bucket rate limiter for LLM and embedding API calls
- **added** — Add Multimodal LLMReranker
- **added** — Add optional embed_model to SemanticDoubleMergingSplitterNodeParser
- **added** — Add SlidingWindowRateLimiter for strict per-minute caps
- **added** — Extend vector store metadata filters
- **added** — Add Neo4j user agent
- **added** — Add apoc_sample parameter for large database schema introspection in Neo4j
- **added** — Add User-Agent header for Anthropic API calls
- **added** — Add ModelsLab LLM integration
- **added** — Add gpt-5-chat support in OpenAI
- **added** — Support reasoning_content in OpenAI Chat Completions
- **fixed** — Fix chonkie initialization documentation
- **fixed** — Pass tool_choice through FunctionCallingProgram
- **fixed** — Preserve doc_id in legacy_json_to_doc
- **fixed** — Fix async retry backoff to avoid blocking event loop
- **fixed** — Fix additionalProperties in auto-generated KG schema models
- **fixed** — Respect db_schema when custom async_engine is provided
- **fixed** — Replace blocking run_async_tasks with asyncio.gather
- **fixed** — Preserve docstore_strategy across pipeline runs when no vector store is attached
- **fixed** — Fix FunctionTool not respecting pydantic Field defaults
- **fixed** — Fix MarkdownElementNodeParser to extract code blocks
- **fixed** — Pass system_prompt to ReActChatFormatter template
- **fixed** — Fix partial-failure handling in SubQuestionQueryEngine
- **fixed** — Add bounds check to prevent infinite loop in ChatMemoryBuffer.get()
- **fixed** — Ensure streaming flag reset on exception in CondenseQuestionChatEngine
- **fixed** — Pass through run id correctly
- **fixed** — Raise ValueError when 'model' is passed instead of 'model_name' in BedrockEmbedding
- **fixed** — Respect Retry-After header in OpenAI retry decorator
- **fixed** — Improve handling of reasoningContent in responses from Bedrock Converse requests
- **fixed** — Forward allow_parallel_tool_calls for OpenAI chat completions
- **fixed** — Use constrained decoding for OpenAIResponses structured_predict
- **fixed** — Apply cache_control only to last block to respect Anthropic's 4-block limit
- **fixed** — Properly manage async client lifecycle in Azure Inference to prevent unclosed sessions
- **deprecated** — Deprecate asyncio_module in favour of get_asyncio_module
- **security** — Add RestrictedUnpickler to SimpleObjectNodeMapping to prevent insecure deserialization
- **changed** — OTEL instrumentation enhancements

#### Release Notes

##### [2026-03-10]

###### llama-index-core [0.14.16]

- Add token-bucket rate limiter for LLM and embedding API calls ([#20712](https://github.com/run-llama/llama_index/pull/20712))
- Fix/20706 chonkie init doc ([#20713](https://github.com/run-llama/llama_index/pull/20713))
- fix: pass tool_choice through FunctionCallingProgram ([#20740](https://github.com/run-llama/llama_index/pull/20740))
- feat: Multimodal LLMReranker ([#20743](https://github.com/run-llama/llama_index/pull/20743))
- feat: add optional embed_model to SemanticDoubleMergingSplitterNodeParser ([#20748](https://github.com/run-llama/llama_index/pull/20748))
- fix(core): preserve doc_id in legacy_json_to_doc ([#20750](https://github.com/run-llama/llama_index/pull/20750))
- fix: async retry backoff to avoid blocking event loop ([#20764](https://github.com/run-llama/llama_index/pull/20764))
- Fix additionalProperties in auto-generated KG schema models ([#20768](https://github.com/run-llama/llama_index/pull/20768))
- fix: respect db_schema when custom async_engine is provided ([#20779](https://github.com/run-llama/llama_index/pull/20779))
- fix(core): replace blocking `run_async_tasks` with `asyncio.gather` ([#20795](https://github.com/run-llama/llama_index/pull/20795))
- feat(rate_limiter): add SlidingWindowRateLimiter for strict per-minute caps ([#20799](https://github.com/run-llama/llama_index/pull/20799))
- fix(core): preserve `docstore_strategy` across pipeline runs when no vector store is attached ([#20824](https://github.com/run-llama/llama_index/pull/20824))
- Fix FunctionTool not respecting pydantic Field defaults ([#20839](https://github.com/run-llama/llama_index/pull/20839))
- Fix MarkdownElementNodeParser to extract code blocks ([#20840](https://github.com/run-llama/llama_index/pull/20840))
- security: add RestrictedUnpickler to SimpleObjectNodeMapping (CWE-502) ([#20857](https://github.com/run-llama/llama_index/pull/20857))
- feat: extend vector store metadata filters ([#20861](https://github.com/run-llama/llama_index/pull/20861))
- fix(react): pass system_prompt to ReActChatFormatter template ([#20873](https://github.com/run-llama/llama_index/pull/20873))
- refactor: deprecate asyncio_module in favour of get_asyncio_module ([#20902](https://github.com/run-llama/llama_index/pull/20902))
- fix(core): partial-failure handling in SubQuestionQueryEngine ([#20905](https://github.com/run-llama/llama_index/pull/20905))
- fix: add bounds check to prevent infinite loop in ChatMemoryBuffer.get() ([#20914](https://github.com/run-llama/llama_index/pull/20914))
- fix: ensure streaming flag reset on exception in CondenseQuestionChatEngine ([#20915](https://github.com/run-llama/llama_index/pull/20915))
- fix: pass through run id correctly ([#20928](https://github.com/run-llama/llama_index/pull/20928))

###### llama-index-embeddings-bedrock [0.7.4]

- fix: raise ValueError when 'model' is passed instead of 'model_name' in BedrockEmbedding ([#20836](https://github.com/run-llama/llama_index/pull/20836))

###### llama-index-embeddings-openai [0.5.2]

- Respect Retry-After header in OpenAI retry decorator ([#20813](https://github.com/run-llama/llama_index/pull/20813))

###### llama-index-embeddings-upstage [0.5.1]

- chore(deps): bump the uv group across 47 directories with 3 updates ([#20793](https://github.com/run-llama/llama_index/pull/20793))

###### llama-index-graph-stores-neo4j [0.6.0]

- Add Neo4j user agent ([#20827](https://github.com/run-llama/llama_index/pull/20827))
- feat(neo4j): add apoc_sample parameter for large database schema introspection ([#20859](https://github.com/run-llama/llama_index/pull/20859))

###### llama-index-instrumentation [0.4.3]

- otel instrumentation enhancements ([#20816](https://github.com/run-llama/llama_index/pull/20816))

###### llama-index-llms-anthropic [0.10.11]

- Add User-Agent header for Anthropic API calls ([#20771](https://github.com/run-llama/llama_index/pull/20771))
- fix: apply cache_control only to

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.16]_

### v0.14.15

- Date: 2026-02-18
- Version: v0.14.15
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.15
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.15

- **added** — AgentMesh integration providing a trust layer for LlamaIndex agents
- **added** — Support for basic operations on multimodal types
- **added** — Recursive LLM type support
- **added** — Multimodal template variable formatting
- **added** — Multimodal prompt templates
- **added** — Multimodal chat prompt helper
- **added** — Retry and error handling to BaseExtractor
- **added** — Support for Claude Sonnet 4.6 in bedrock-converse
- **added** — Support for new OCI DataScience endpoint /predictWithStream for streaming use case
- **added** — Custom span processor support in observability-otel
- **added** — Sandbox LLM-generated code execution in EvaporateExtractor
- **added** — LayoutIR reader integration
- **added** — Pagination support for Microsoft Graph API calls in SharePoint reader
- **added** — Partial parameter propagation to get_tools_from_mcp utils
- **changed** — GitHubRepoReader enhanced with selective file fetching and deduplication
- **changed** — Observability-otel now uses llama-index-instrumentation instead of llama-index-core
- **changed** — MistralAI LLM integration updated for Azure SDK compatibility
- **fixed** — Remove redundant metadata_seperator field from TextNode
- **fixed** — Ensure thinking_delta is populated in all chat modes for bedrock-converse
- **fixed** — Ensure at least one message/content block is returned by the old memory
- **fixed** — Replace mutable default argument in Bitbucket reader load_all_file_paths
- **fixed** — LayoutIR reader hotfix for output_dir crash and Block extraction
- **fixed** — Restrict LayoutIR requires-python to >=3.12 to match layoutir dependency
- **fixed** — Update WhatsAppChatLoader to retrieve DataFrame in pandas format
- **fixed** — Replace eval() with json.loads in FaissMapVectorStore persistence
- **fixed** — Remove ORM Collection mix-usage with MilvusClient in Milvus vector store
- **removed** — Persistent_connection parameter support from IBM embeddings and LLM integrations

#### Release Notes

##### [2026-02-18]

###### llama-index-agent-agentmesh [0.1.0]

- [Integration] AgentMesh: Trust Layer for LlamaIndex Agents ([#20644](https://github.com/run-llama/llama_index/pull/20644))

###### llama-index-core [0.14.15]

- Support basic operations for multimodal types ([#20640](https://github.com/run-llama/llama_index/pull/20640))
- Feat recursive llm type support ([#20642](https://github.com/run-llama/llama_index/pull/20642))
- fix: remove redundant metadata_seperator field from TextNode ([#20649](https://github.com/run-llama/llama_index/pull/20649))
- fix(tests): update mock prompt type in mock_prompts.py ([#20661](https://github.com/run-llama/llama_index/pull/20661))
- Feat multimodal template var formatting ([#20682](https://github.com/run-llama/llama_index/pull/20682))
- Feat multimodal prompt templates ([#20683](https://github.com/run-llama/llama_index/pull/20683))
- Feat multimodal chat prompt helper ([#20684](https://github.com/run-llama/llama_index/pull/20684))
- Add retry and error handling to BaseExtractor ([#20693](https://github.com/run-llama/llama_index/pull/20693))
- ensure at least one message/content block is returned by the old memory ([#20729](https://github.com/run-llama/llama_index/pull/20729))

###### llama-index-embeddings-ibm [0.6.0.post1]

- chore: Remove persistent_connection parameter support, update ([#20714](https://github.com/run-llama/llama_index/pull/20714))
- docs: Update IBM docs ([#20718](https://github.com/run-llama/llama_index/pull/20718))

###### llama-index-llms-anthropic [0.10.9]

- Sonnet 4-6 addition ([#20723](https://github.com/run-llama/llama_index/pull/20723))

###### llama-index-llms-bedrock-converse [0.12.10]

- fix(bedrock-converse): ensure thinking_delta is populated in all chat modes ([#20664](https://github.com/run-llama/llama_index/pull/20664))
- feat(bedrock-converse): Add support for Claude Sonnet 4.6 ([#20726](https://github.com/run-llama/llama_index/pull/20726))

###### llama-index-llms-ibm [0.7.0.post1]

- chore: Remove persistent_connection parameter support, update ([#20714](https://github.com/run-llama/llama_index/pull/20714))
- docs: Update IBM docs ([#20718](https://github.com/run-llama/llama_index/pull/20718))

###### llama-index-llms-mistralai [0.10.0]

- Rrubini/mistral azure sdk ([#20668](https://github.com/run-llama/llama_index/pull/20668))

###### llama-index-llms-oci-data-science [1.0.0]

- Add support for new OCI DataScience endpoint /predictWithStream for streaming use case ([#20545](https://github.com/run-llama/llama_index/pull/20545))

###### llama-index-observability-otel [0.3.0]

- improve otel data serialization by flattening dicts ([#20719](https://github.com/run-llama/llama_index/pull/20719))
- feat: support custom span processor; refactor: use llama-index-instrumentation instead of llama-index-core ([#20732](https://github.com/run-llama/llama_index/pull/20732))

###### llama-index-program-evaporate [0.5.2]

- Sandbox LLM-generated code execution in EvaporateExtractor ([#20676](https://github.com/run-llama/llama_index/pull/20676))

###### llama-index-readers-bitbucket [0.4.2]

- fix: replace mutable default argument in load_all_file_paths ([#20698](https://github.com/run-llama/llama_index/pull/20698))

###### llama-index-readers-github [0.10.0]

- feat: Enhance GitHubRepoReader with selective file fetching and deduplication (Issue #20471) ([#20550](https://github.com/run-llama/llama_index/pull/20550))

###### llama-index-readers-layoutir [0.1.1]

- feat: Add LayoutIR reader integration ([#20708](https://github.com/run-llama/llama_index/pull/20708))
- fix(layoutir): hotfix for output_dir crash and Block extraction (#20708 follow-up) ([#20715](https://github.com/run-llama/llama_index/pull/20715))
- fix(layoutir): restrict requires-python to >=3.12 to match layoutir dependency ([#20733](https://github.com/run-llama/llama_index/pull/20733))

###### llama-index-readers-microsoft-sharepoint [0.8.0]

- Add pagination support for Micr

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.15]_

### v0.14.14

- Date: 2026-02-10
- Version: v0.14.14
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.14
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.14

- **fixed** — catch pydantic ValidationError in VectorStoreQueryOutputParser
- **fixed** — distinguish empty string from None in MediaResource.hash
- **added** — add Langchain 1.x support
- **fixed** — fix DeprecationWarning: 'asyncio.iscoroutinefunction' is deprecated
- **fixed** — fallback to bundled nltk cache if env var missing
- **added** — add TokenBudgetHandler for cost governance
- **fixed** — handle edge case in truncate_text function
- **fixed** — fix Thread passing None when target is None instead of copy_context().run
- **changed** — compatibility for workflows context changes
- **fixed** — handle empty LLM responses with retry logic in agent
- **fixed** — add show_progress parameter to run_transformations to prevent unexpected keyword argument error
- **fixed** — fix potential crashes and improve security defaults in core components
- **fixed** — add retry logic with tenacity to Cohere embeddings
- **added** — add client headers to Gemini API requests
- **fixed** — fix MENTIONS relationship creation with triplet_source_id in FalkorDB
- **changed** — update cacheable Anthropic models
- **added** — add support for Opus 4.6
- **fixed** — fix bedrock converse empty tool config issue
- **changed** — improve bedrock converse retry handling
- **added** — add support for adaptive thinking in Bedrock

#### Release Notes

##### [2026-02-10]

###### llama-index-callbacks-wandb [0.4.2]

- Fix potential crashes and improve security defaults in core components ([#20610](https://github.com/run-llama/llama_index/pull/20610))

###### llama-index-core [0.14.14]

- fix: catch pydantic ValidationError in VectorStoreQueryOutputParser ([#20450](https://github.com/run-llama/llama_index/pull/20450))
- fix: distinguish empty string from None in MediaResource.hash ([#20451](https://github.com/run-llama/llama_index/pull/20451))
- Langchain1.x support ([#20472](https://github.com/run-llama/llama_index/pull/20472))
- Fix DeprecationWarning: 'asyncio.iscoroutinefunction' is deprecated ([#20517](https://github.com/run-llama/llama_index/pull/20517))
- fix(core): fallback to bundled nltk cache if env var missing ([#20528](https://github.com/run-llama/llama_index/pull/20528))
- feat(callbacks): add TokenBudgetHandler for cost governance ([#20546](https://github.com/run-llama/llama_index/pull/20546))
- fix(core):handled a edge case in truncate_text function ([#20551](https://github.com/run-llama/llama_index/pull/20551))
- fix(core):fix in types Thread passing None when target is None instead of copy_context().run ([#20553](https://github.com/run-llama/llama_index/pull/20553))
- chore: bump llama-index lockfile, and minor test tweaks ([#20556](https://github.com/run-llama/llama_index/pull/20556))
- Compatibility for workflows context changes ([#20557](https://github.com/run-llama/llama_index/pull/20557))
- test(core): fix cache dir path test for Windows compatibility ([#20566](https://github.com/run-llama/llama_index/pull/20566))
- fix(tests): enforce utf-8 encoding in json reader tests for windows compatibility ([#20576](https://github.com/run-llama/llama_index/pull/20576))
- Fix BM25Retriever mapping in upgrade tool / 修复升级工具中的 BM25Retriever 映射 ([#20582](https://github.com/run-llama/llama_index/pull/20582))
- fix(agent): handle empty LLM responses with retry logic and add test cases ([#20596](https://github.com/run-llama/llama_index/pull/20596))
- fix: add show_progress parameter to run_transformations to prevent unexpected keyword argument error ([#20608](https://github.com/run-llama/llama_index/pull/20608))
- Fix potential crashes and improve security defaults in core components ([#20610](https://github.com/run-llama/llama_index/pull/20610))
- Add core 3.14 tests ([#20619](https://github.com/run-llama/llama_index/pull/20619))

###### llama-index-embeddings-cohere [0.7.0]

- fix(embeddings-cohere): add retry logic with tenacity ([#20592](https://github.com/run-llama/llama_index/pull/20592))

###### llama-index-embeddings-google-genai [0.3.2]

- Add client headers to Gemini API requests ([#20519](https://github.com/run-llama/llama_index/pull/20519))

###### llama-index-embeddings-siliconflow [0.3.2]

- Fix DeprecationWarning: 'asyncio.iscoroutinefunction' is deprecated ([#20517](https://github.com/run-llama/llama_index/pull/20517))

###### llama-index-embeddings-upstage [0.5.1]

- chore(deps): bump the uv group across 4 directories with 4 updates ([#20531](https://github.com/run-llama/llama_index/pull/20531))

###### llama-index-graph-stores-falkordb [0.4.2]

- fix(falkordb): Fix MENTIONS relationship creation with triplet_source_id ([#20650](https://github.com/run-llama/llama_index/pull/20650))

###### llama-index-llms-anthropic [0.10.8]

- chore: Update cacheable Anthropic models ([#20581](https://github.com/run-llama/llama_index/pull/20581))
- chore: add support for opus 4.6 ([#20635](https://github.com/run-llama/llama_index/pull/20635))

###### llama-index-llms-bedrock-converse [0.12.8]

- fix bedrock converse empty tool config issue ([#20571](https://github.com/run-llama/llama_index/pull/20571))
- fix(llms-bedrock-converse): improve bedrock converse retry handling ([#20590](https://github.com/run-llama/llama_index/pull/20590))
- feat(bedrock-converse): Add support for Claude Opus 4.6 ([#20637](https://github.com/run-llama/llama_index/pull/20637

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.14]_

### v0.14.13

- Date: 2026-01-21
- Version: v0.14.13
- Original notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.13
- Permalink: https://whatsnew.fyi/product/llamaindex/releases/v0.14.13

- **added** — Add early_stopping_method parameter to agent workflows
- **added** — Add token-based code splitting support to CodeSplitter
- **added** — Add RayIngestionPipeline integration for distributed data ingestion
- **added** — Add multi-modal version of Condensed Conversation & Context
- **changed** — Replace ChatMemoryBuffer with Memory
- **fixed** — Raise value error when input is empty list in mean_agg instead of returning float
- **fixed** — The classmethod of ReActChatFormatter should use cls instead of the class name
- **added** — Add configurable empty response message to synthesizers
- **changed** — Enable use of ARNs for Bedrock Embedding Models
- **changed** — Improved Ollama batch embedding
- **added** — Add voyage-4 models
- **changed** — Enhance structured predict methods for anthropic
- **fixed** — Preserve input_tokens in Anthropic stream_chat responses
- **added** — Add Apertis LLM integration
- **changed** — Remove extraneous thinking_delta kwarg from ChatMessage in bedrock-converse
- **deprecated** — Deprecate llama-index-llms-gemini
- **fixed** — Sanitize OpenAI structured output JSON schema name for generic Pydantic models
- **added** — Add provider routing support to OpenRouter
- **security** — Remove exposed OpenAI API keys from notebook outputs
- **added** — Add HuggingFace datasets reader integration

#### Release Notes

##### [2026-01-21]

###### llama-index-core [0.14.13]

- feat: add early_stopping_method parameter to agent workflows ([#20389](https://github.com/run-llama/llama_index/pull/20389))
- feat: Add token-based code splitting support to CodeSplitter ([#20438](https://github.com/run-llama/llama_index/pull/20438))
- Add RayIngestionPipeline integration for distributed data ingestion ([#20443](https://github.com/run-llama/llama_index/pull/20443))
- Added the multi-modal version of the Condensed Conversation & Context… ([#20446](https://github.com/run-llama/llama_index/pull/20446))
- Replace ChatMemoryBuffer with Memory ([#20458](https://github.com/run-llama/llama_index/pull/20458))
- fix(bug):Raise value error on when input is empty list in mean_agg instead of returning float ([#20466](https://github.com/run-llama/llama_index/pull/20466))
- fix: The classmethod of ReActChatFormatter should use cls instead of the class name ([#20475](https://github.com/run-llama/llama_index/pull/20475))
- feat: add configurable empty response message to synthesizers ([#20503](https://github.com/run-llama/llama_index/pull/20503))

###### llama-index-embeddings-bedrock [0.7.3]

- Enable use of ARNs for Bedrock Embedding Models ([#20435](https://github.com/run-llama/llama_index/pull/20435))

###### llama-index-embeddings-ollama [0.8.6]

- Improved Ollama batch embedding ([#20447](https://github.com/run-llama/llama_index/pull/20447))

###### llama-index-embeddings-voyageai [0.5.3]

- Adding voyage-4 models ([#20497](https://github.com/run-llama/llama_index/pull/20497))

###### llama-index-ingestion-ray [0.1.0]

- Add RayIngestionPipeline integration for distributed data ingestion ([#20443](https://github.com/run-llama/llama_index/pull/20443))

###### llama-index-llms-anthropic [0.10.6]

- feat: enhance structured predict methods for anthropic ([#20440](https://github.com/run-llama/llama_index/pull/20440))
- fix: preserve input_tokens in Anthropic stream_chat responses ([#20512](https://github.com/run-llama/llama_index/pull/20512))

###### llama-index-llms-apertis [0.1.0]

- Add Apertis LLM integration with example notebook ([#20436](https://github.com/run-llama/llama_index/pull/20436))

###### llama-index-llms-bedrock-converse [0.12.4]

- chore(bedrock-converse): Remove extraneous thinking_delta kwarg from ChatMessage ([#20455](https://github.com/run-llama/llama_index/pull/20455))

###### llama-index-llms-gemini [0.6.2]

- chore: deprecate llama-index-llms-gemini ([#20511](https://github.com/run-llama/llama_index/pull/20511))

###### llama-index-llms-openai [0.6.13]

- Sanitize OpenAI structured output JSON schema name for generic Pydantic models ([#20452](https://github.com/run-llama/llama_index/pull/20452))
- chore: vbump openai ([#20482](https://github.com/run-llama/llama_index/pull/20482))

###### llama-index-llms-openrouter [0.4.3]

- Feature/openrouter provider routing support ([#20431](https://github.com/run-llama/llama_index/pull/20431))

###### llama-index-packs-recursive-retriever [0.7.1]

- security: remove exposed OpenAI API keys from notebook outputs ([#20474](https://github.com/run-llama/llama_index/pull/20474))

###### llama-index-packs-sentence-window-retriever [0.5.1]

- security: remove exposed OpenAI API keys from notebook outputs ([#20474](https://github.com/run-llama/llama_index/pull/20474))

###### llama-index-readers-datasets [0.1.0]

- Add HuggingFace datasets reader integration ([#20468](https://github.com/run-llama/llama_index/pull/20468))

###### llama-index-readers-patentsview [1.0.0]

- Patentsview reader api changes ([#20481](https://github.com/run-llama/llama_index/pull/20481))

###### llama-index-retrievers-you [1.0.0]

- Revamp YouRetriever integration ([#20493](https://github.com/run-llama/llama_index/pull/20493))

###### llama-index-tools-parallel-web-systems [0.1.0]

- feat: added Parallel Web System tools ([#20442](https://github.com/run-llama/llama_index/pull/20442))

###### llama-index-vector-store

_[Truncated at 4000 characters — full notes: https://github.com/run-llama/llama_index/releases/tag/v0.14.13]_