The daily firehose — everything the toolchain shipped today, already filtered.
// HOW THIS ISSUE IS MADE
We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.
AutoGPT v0.3.0 adds a plugin system, automatic prompt generation, self-feedback mode, running-cost awareness, and authenticated Milvus memory backends.
└──▷ GET THIS VERSION
$ git clone --branch v0.3.0 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout v0.3.0
└──▷ HOW TO FIND IT
Trigger self-feedback to let the agent critique and refine its own plan mid-run.
📍When AutoGPT prompts for input, press: S
›Adds <reason> argument to the do_nothing command, allowing the agent to record why it chose to take no action.
›Supports secure and authenticated Milvus memory backends for production deployments.
›Introduces a third-party plugin system so developers can extend AutoGPT with new commands and integrations (note: plugin interface is marked unstable and may change in v0.3.1 and v0.4.0).
›Adds automatic initial prompt generation — describe a goal in plain language and AutoGPT generates its own structured prompt.
›Adds self-feedback mode: pressing S at the input prompt triggers the AI to reflect on and revise its own reasoning and plans.
+6 moreshow less
›Adds running-cost awareness so AutoGPT tracks and surfaces its accumulated API spend during a session.
›Adds a workspace abstraction to isolate and manage the agent's file operations.
›Adds OS info into the initial prompt so the agent is aware of the host environment.
›Includes memory challenge benchmarks (levels 1–10) in the test suite to objectively measure memory system improvements.
›Maintains a running summary of prior interactions after each step to improve long-session memory management.
›Handles API timeouts gracefully instead of crashing.
└──▷ BREAKING ON UPGRADE
!The blacklist and whitelist configuration terms are renamed to denylist and allowlist respectively; existing configs using the old terms will need to be updated.
Haystack v1.17 adds ConversationalAgent with memory, Anthropic and Cohere LLM support, Weaviate auth, and streaming for HF Inference Endpoints.
└──▷ GET THIS VERSION
$ git clone --branch v1.17.0 https://github.com/deepset-ai/haystack.git
# already have the repo? check out this version:$ git checkout v1.17.0
└──▷ USE IT
Build a chat application with summarized memory to stay within token limits.
python
from haystack.agents.memory import ConversationalSummaryMemory
from haystack.agents import ConversationalAgent
summary_memory = ConversationalSummaryMemory(prompt_node=prompt_node)
agent = ConversationalAgent(prompt_node=prompt_node, memory=summary_memory)
response = agent.run(user_input="What are the main causes of climate change?")
Override generation parameters per pipeline run without changing the PromptNode definition.
›Adds ConversationalAgent class for building chat applications, accepting a PromptNode and an optional memory argument for conversation history injection.
›Adds ConversationSummaryMemory (also referenced as ConversationalSummaryMemory) to condense chat history before injecting into the prompt, keeping usage within model token limits.
›Adds AnthropicInvocationLayer to support claude models from Anthropic as a PromptNode backend.
›Adds CohereInvocationLayer to support command models from Cohere as a PromptNode backend.
›Adds AuthBearerToken and AuthClientCredentials authentication options to WeaviateDocumentStore.
+7 moreshow less
›Adds max_tokens parameter to BaseGenerator params, exposing token-limit control across generator implementations.
›Adds streaming support to HFInferenceEndpointInvocationLayer for token-by-token output from Hugging Face Inference Endpoints.
›Adds streaming support to the HF local runtime invocation layer.
›Enables passing generation_kwargs to PromptNode at pipeline.run() time, allowing per-run overrides of generation parameters.
›Adds BLIP model support to TransformersImageToText component.
›Adds Google API as a search engine provider option.
›Introduces generalimport to defer missing-dependency errors from import time to actual usage time, reducing mandatory dependencies for a base pip install farm-haystack.
└──▷ BREAKING ON UPGRADE
!MilvusDocumentStore is removed from core Haystack; it must now be installed separately from the haystack-extras repo.
!BaseKnowledgeGraph is removed from the library.
!The PDFToTextOCRConverter node is removed.
!Schema objects' to_dict, from_dict, to_json, and from_json methods have been updated to handle Dataframes, which may change serialization behavior for existing code.
Analyse multiple CSV files at once using the updated multi-CSV agent toolkit.
python
from langchain.agents import create_csv_agent
from langchain.llms import OpenAI
agent = create_csv_agent(
OpenAI(temperature=0),
['users.csv', 'events.csv'],
verbose=True
)
agent.run('Which user triggered the most events?')
›Adds visible_only and strict_mode options to ClickTool for finer control over browser automation interactions.
›Adds pipeline_kwargs support to HuggingFacePipeline.from_model_id for passing arbitrary pipeline arguments at construction time.
›Adds support for the BigQuery SQL dialect in the SQL database integration.
›Adds C Transformers integration for running GGML-format local models via a new LLM wrapper.
›Adds Momento as both a standard LLM cache provider and a chat message history backend.
+4 moreshow less
›Adds a Twilio tool, enabling agents to send messages via Twilio.
›Adds an LLM wrapper for Databricks, enabling LangChain chains and agents to call Databricks-hosted models.
›Adds a proxy configuration option for the OpenAI API client.
›Adds multi-CSV and multi-DataFrame support to the CSV and DataFrame agent toolkits.
LangChain v0.0.180 adds ModelScope and Vertex AI integrations, TF-IDF retriever, BibTeX loader, MiniMax embeddings, and more new loaders and capabilities.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.180 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.180
└──▷ USE IT
Quickly build a sparse retriever from a set of documents when you have no vector DB available.
python
from langchain.retrievers import TFIDFRetriever
retriever = TFIDFRetriever.from_documents(docs)
results = retriever.get_relevant_documents('what is the capital of France?')
›Adds TFIDFRetriever for sparse retrieval over document collections without a vector database.
›Adds BibtexLoader and a BibTeX-backed retriever for loading and retrieving academic references from .bib files.
›Adds MiniMaxEmbeddings for generating embeddings via the MiniMax API.
›Adds IuguLoader document loader for ingesting Iugu financial data.
›Adds JoplinLoader document loader for loading notes from a Joplin instance.
+9 moreshow less
›Adds ModelScope LLM integration (Harrison/modelscope) for accessing ModelScope-hosted models.
›Adds Google Vertex AI LLM integration (Harrison/vertex) for accessing Vertex AI language models.
›Adds async from_text() method to GraphIndexCreator for non-blocking knowledge graph construction.
›Adds status subcommand to the langchain plus CLI to check LangChain Plus server status.
›Adds Delete Session method to conversation session management.
›Adds option to pass an OpenAI API key directly to the langchain plus CLI command.
›Allows specifying a custom ID when adding documents to a FAISS vectorstore.
›Allows ReadTheDocsLoader to accept a custom HTML tag for more flexible documentation ingestion.
›Changes default GoogleDriveLoader behavior to skip trashed files.
└──▷ BREAKING ON UPGRADE
!The default behavior of GoogleDriveLoader changes: trashed files are no longer loaded. Pipelines relying on trashed-file ingestion will silently stop receiving those documents.
›Adds Rebuff integration for prompt injection detection and defense in LLM pipelines.
›Adds TelegramChatLoader for loading Telegram chat history as documents.
›Adds DocugamiLoader for loading documents from Docugami.
+5 moreshow less
›Adds PDFPlumberLoader (using BaseBlobParser) for PDF ingestion via the pdfplumber library.
›Adds streaming output support to HuggingFaceTextgenInference LLM class.
›Adds support for loading sitemaps from local files in the sitemap loader.
›Improves YoutubeLoader video ID extraction using built-in URL parsing instead of regex, broadening supported URL formats.
›Adds environment info to LangChain runs for better observability and debugging context.
└──▷ BREAKING ON UPGRADE
!The openai_api_version parameter is no longer set by default in the OpenAI integration; setups relying on a default value must now supply it explicitly.
LangChain v0.0.155 adds Google PaLM models, ConstitutionalChain, Cohere reranker, SQLite chat history, Unstructured API loaders, and a Structured Chat Agent.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.155 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.155
└──▷ USE IT
Use Google PaLM as a drop-in LLM backend for any LangChain chain.
python
from langchain.llms import GooglePalm
llm = GooglePalm(google_api_key='<your-api-key>')
print(llm('Explain zero-trust architecture in one sentence.'))
Persist conversation history to SQLite so it survives process restarts.
python
from langchain.memory import SQLiteChatMessageHistory
history = SQLiteChatMessageHistory(session_id='user-123', connection_string='sqlite:///chat.db')
Wrap a chain in ConstitutionalChain to automatically critique and revise unsafe or low-quality outputs.
python
from langchain.chains import ConstitutionalChain, LLMChain
from langchain.chains.constitutional_ai.models import ConstitutionalPrinciple
from langchain.llms import OpenAI
llm = OpenAI()
base_chain = LLMChain(llm=llm, prompt=my_prompt)
constitutional_chain = ConstitutionalChain.from_llm(
llm=llm,
chain=base_chain,
constitutional_principles=[
ConstitutionalPrinciple(
critique_request='Does the response contain harmful content?',
revision_request='Rewrite it to be safe and helpful.'
)
]
)
print(constitutional_chain.run('How do I pick a lock?'))
›Exports StructuredTool at the /tools module path for easier importing.
›Adds SQLiteChatMessageHistory for persistent SQLite-backed conversation memory.
›Adds ChatModel, LLM, and Embeddings classes for Google's PaLM APIs.
›Adds encode_kwargs support to HuggingFace embeddings for finer control over encoding.
›Adds Unstructured API loaders for document ingestion via the Unstructured API.
+16 moreshow less
›Adds ConstitutionalChain for self-critique and revision of LLM outputs.
›Adds CombinedMemory to compose multiple memory backends together.
›Adds a Structured Chat Agent capable of handling structured tool inputs.
›Adds a Cohere reranker for relevance-based document reordering in retrieval pipelines.
›Adds a minimal file system blob loader for loading files as blobs.
›Adds blockwise sitemap loader for large sitemap processing.
›Adds async support to LLMChainExtractor.
›Adds connection string authentication support to the Cosmos DB integration.
›Adds a Modern Treasury API integration.
›Adds Spreedly API integration.
›Adds from_documents class method for constructing vectorstores directly from documents.
›Adds agent_executor_kwargs to allow passing additional keyword arguments to AgentExecutor.
›Adds relevancy score support to similarity search results.
›Adds multi-agent simulation with environment example using GymnasiumAgent.
›Counts tokens instead of characters in AutoGPT prompt construction for more accurate context management.
›Makes ddg-search available via __init__ for simpler tool loading.
└──▷ BREAKING ON UPGRADE
!GPT4All integration now requires PyGPT4All instead of the previous backend — existing GPT4All setups will break on upgrade without migrating to PyGPT4All.
›Adds token probability display to --debugmode: for every generated token, the console shows probabilities of up to 4 alternative tokens after all samplers are applied, enabling sampler configuration testing and model confidence analysis.
›Adds --debugmode display of input/context contents and their token IDs (note: slight performance hit; off by default).
›Adds --unbantokens flag to enable EOS stop tokens across all model types; the [ token is also no longer banned by default.
›Adds the Top-A sampler, a Kobold-exclusive implementation not present in upstream llama.cpp, which reduces randomness proportionally to the squared softmax probability of the most probable token (set to 0 to disable).
›Adds support for Starcoder and Starcoder Chat models.
7 more releases in this issue
· 2023-05-01 → 2023-05-27
Set a fixed sampler seed via the KoboldAI /generate API for reproducible outputs.
$ curl -X POST http://localhost:5001/api/v1/generate -H 'Content-Type: application/json' -d '{"prompt": "Once upon a time", "sampler_seed": 42}'
›Adds Failsafe mode via --noavx2 --noblas --nommap flags, disabling all CPU intrinsics (AVX, SSE) to run on ancient hardware including Windows 7.
›Adds sampler_seed parameter support to the /generate API endpoint for reproducible generation.
›Adds ?streamamount=[value] URL parameter to Kobold Lite UI for controlling variable streaming lengths (default: 8 tokens) when launched with --stream.
›Expands --debugmode console logging to display context token contents.
›Adds drag-and-drop file load functionality to Kobold Lite UI.
+2 moreshow less
›Greatly improved markdown rendering support in Kobold Lite UI.
KoboldCpp v1.24 adds GGJT v3 quantization support (q4_0, q4_1, q8_0) and new Kobold Lite UI toggles.
└──▷ GET THIS VERSION
$ git clone --branch v1.24 https://github.com/LostRuins/koboldcpp.git
# already have the repo? check out this version:$ git checkout v1.24
›Supports the new GGJT v3 quantization format, including q4_0, q4_1, and q8_0 quantization types, while retaining backwards compatibility with all historical GGML formats (GGML, GGHF, GGJT v1, v2, v3).
›Adds a toggle in Kobold Lite to avoid inserting newlines in Instruct mode, useful for Pygmalion and OpenAssistant based instruct models.
›Adds a toggle in Kobold Lite to enable basic markdown rendering in Instruct mode (off by default).
›Provides an alternative CUDA build (via Henky) for this version, enabling access to the latest GGJT v3 quantizations for CUDA users on LLAMA-based models.
›Adds experimental OpenCL GPU offloading via CLBlast using --useclblast combined with --gpulayers <n> to select the number of layers to offload; works on all GPUs for new quantization formats of LLAMA models.
›Extends GPU offloading support to q8 quantization formats after pulling the q8 dequant kernel fix.
›Adds support for new quantization formats for GPT-2, GPT-J, and GPT-NeoX models.
›Adds --usemirostat [type] [tau] [eta] flag to enable mirostat sampling on all model types, replacing normal stochastic samplers; e.g. --usemirostat 2 5.0 0.1 for mirostat type 2.
›Adds --forceversion [ver] flag to override automatic model file format detection when it fails, e.g. 401 for GPTNeoX-Type2.
›Adds --blasthreads flag to set a separate thread count when CLBlast is active, defaulting to the value of --threads if not specified.
›Expands RWKV support to include all new RWKV quantizations, with q5_1 delivering significantly faster inference than fp16 at similar quality.
›Includes an experimental Windows 7-compatible .exe build for this release.
└──▷ BREAKING ON UPGRADE
!RWKV Q4_1_O quantization is no longer supported following the upstream change.
Keep a large model pinned in RAM on Apple M1 to avoid swapping and reduce generation latency.
$ koboldcpp.exe --model mymodel.bin --usemlock
›Adds --highpriority CLI flag to raise the process CPU priority, potentially reducing generation latency.
›Adds --usemlock CLI parameter to pin the model in RAM, targeting Apple M1 users.
›Adds Group Conversations to Kobold Lite chat mode: specify multiple chat opponents delimited with ||$|| (up to 10 custom stopping sequences) so the AI replies as different participants; works best with Multiline Replies disabled and on LLAMA-based models.
›Adds a new built-in scenario Class Reunion in Kobold Lite to demonstrate group chat functionality.
KoboldCpp v1.17 adds --unbantokens flag to unban tokens including EOS, required for newer Pygmalion models.
└──▷ GET THIS VERSION
$ git clone --branch v1.17 https://github.com/LostRuins/koboldcpp.git
# already have the repo? check out this version:$ git checkout v1.17
└──▷ TRY IT
Run KoboldCpp with token unbanning enabled to support newer Pygmalion models that require EOS suppression.
$ koboldcpp.exe --unbantokens
›Adds --unbantokens CLI flag to unban tokens during generation, including prevention of EOS token generation, enabling compatibility with newer Pygmalion models.
›Exposes Token Unbanning in the UI, allowing it to be configured without CLI flags.
ONNX Runtime v1.15.0 adds JS and QNN execution providers, on-device training, WebGPU preview, and multi-GPU collective support.
└──▷ GET THIS VERSION
$ git clone --branch v1.15.0 https://github.com/microsoft/onnxruntime.git
# already have the repo? check out this version:$ git checkout v1.15.0
└──▷ TRY IT
Build ONNX Runtime on a POSIX system as root (e.g., inside a Docker container running as root) — required now that root builds are disallowed by default.
›Adds --allow_running_as_root flag to the build command to permit building as root on POSIX systems (default now disallows it).
›Adds ONNX Optional type support in the C# API.
›Adds collective operations to support multi-GPU inferencing.
›Adds Python 3.11 support across onnxruntime CPU, onnxruntime-gpu, onnxruntime-directml, and onnxruntime-training packages (drops 3.7).
›Introduces two new execution providers: JS EP (JavaScript) and QNN EP (Qualcomm Neural Network).
+23 moreshow less
›Adds initial public preview of QNN EP, available as a NuGet package (Microsoft.ML.OnnxRuntime.QNN).
›Adds official support for TensorRT 8.6, including explicit shape profile overrides, TensorRT plugin support via ORT custom ops, and timing cache.
›Adds support for TensorRT options: heuristics, sparsity, optimization level, auxiliary stream, and tactic source selection.
›Adds support for OpenVINO 2023.0 and dynamic shapes for iGPU in OpenVINO EP.
›Adds OpenAI Whisper model support in Azure EP, now also available as a NuGet package.
›Adds DirectML 1.12 support with opset 16–17 coverage in DirectML EP.
›Adds Swift Package Manager package for onnxruntime on mobile.
›Adds NuGet package for onnxruntime-extensions with Android/iOS support for MAUI/Xamarin.
›Adds React Native package for onnxruntime with optional onnxruntime-extensions inclusion.
›Adds built-in pre/post processing for NLP scenarios (classification, question-answering, text-prediction) on mobile.
›Adds built-in pre/post processing for Speech Recognition (Whisper) on mobile.
›Adds built-in post processing for Object Detection (YOLO) including non-max suppression and bounding box drawing on mobile.
›Adds NNAPI kernels for BatchNormalization and LRN; CoreML kernels for Div, Flatten, LeakyRelu, LRN, Mul, Pad, Pow, and Sub.
›Adds [preview] WebGPU support in the Web build.
›Adds official On-Device Training package with APIs and language bindings for C, C++, Python, C#, and Java; packages available for Desktop and Android.
›Adds graph optimizations leveraging label-data sparsity for ORT Training, yielding 4%–15% performance gains on popular Hugging Face models.
›Adds native Windows ARM64 build support using Visual Studio 2022.
›Updates to CUDA 11.8 (source remains compatible with CUDA 11.4 and 12.x).
›Adds a lock-free queue build option for threadpool to improve CPU utilization.
›Adds fused decoder multi-head attention kernel improving GPT and decoder model (T5, Whisper) performance.
›Adds packing mode to improve encoder model performance with inputs of large padding ratio.
›Upgrades DNNL from 2.7.1 to 3.0.
›Adds cutlass as a new dependency for CUDA/TensorRT packages.
└──▷ BREAKING ON UPGRADE
!The onnxruntime_ENABLE_EAGER_MODE CMake option and eager mode code are deleted — builds using this option will fail.
!Dropped support for Windows 8.1 and below.
!In v1.16.0 (announced here): support for iOS 11 and below will be dropped; iOS 12 will be the minimum.
!In v1.16.0 (announced here): support for CentOS 7, Ubuntu 18.04, and Linux distros without glibc >= 2.28 will be dropped.
!In v1.16.0 (announced here): support for GCC <= 9 and Visual Studio 2019 will be dropped.
!In v1.16.0 (announced here): the onnxruntime_DISABLE_ABSEIL build option will be removed.
Triton 2.34.0 adds custom metrics in Python backend, a client plugin API, dynamic instance scaling, and a new --metrics-address flag.
└──▷ GET THIS VERSION
$ git clone --branch v2.34.0 https://github.com/triton-inference-server/server.git
# already have the repo? check out this version:$ git checkout v2.34.0
└──▷ TRY IT
Bind the Prometheus metrics endpoint to a specific internal interface instead of the default 0.0.0.0, useful when you need to restrict metrics exposure on a multi-homed host.
On a high-core-count host where the old default caused resource exhaustion, pin model load parallelism to a safe value while keeping the new lower default as a baseline.
›Adds --metrics-address=<address> CLI option to bind the metrics server to a different address than the default 0.0.0.0.
›Reduces the default number of model load threads from 2*(number of CPU cores) to 4; the --model-load-thread-count CLI option overrides this default.
›Python backend now supports Custom Metrics, letting users define and report counters and gauges via the same interface as the C API.
›Python Triton Client introduces the Triton Client Plugin API (beta) for registering custom plugins that add or modify request headers.
›Adds DLPack Python specification support in the Python backend via pb_utils.Tensor.from_dlpack().
+1 moreshow less
›Improves model instance scaling: when only the instance group changes in a model config, Triton updates instance counts in-place without a full model reload (non-sequence models only).
└──▷ BREAKING ON UPGRADE
!The default model load thread count is reduced from 2*(number of CPU cores) to 4; deployments on large-core systems that relied on the previous default parallelism will need to set --model-load-thread-count explicitly to restore prior behavior.
Zed v0.88.3 adds GPT-4 in-buffer AI assist, a new panel system, and git status toggle.
└──▷ GET THIS VERSION
$ git clone --branch v0.88.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:$ git checkout v0.88.3
└──▷ USE IT
Enable git status indicators in the project panel via settings.
json
{
"project_panel": {
"git_status": true
}
}
›Adds project_panel: { git_status: bool } setting to control whether git status information appears in the project panel.
›Introduces ai: assist command to pass the current buffer to GPT-4 when OPENAI_API_KEY is set in the environment; .zmd files also support cmd-enter to invoke the model.
›Adds MoveToStartOfParagraph and MoveToEndOfParagraph movement commands for paragraph-based vertical navigation.
›Introduces a more flexible and customizable panel system.
›Adds a recently opened file list to the search file dialogue.
4 more releases in this issue
· 2023-05-03 → 2023-05-31
›Adds scrollbar.show setting (values: auto, system, always, never) and scrollbar.git_diff boolean to control scrollbar visibility and git diff markers in the scrollbar.
›Adds git diff locations as visual markers in the editor scrollbar.
›Adds git status colors to files and directories in the project panel.
›Enables jumping to a specific line and optional column from the file finder and zed CLI by appending :<line>:<column> after a filename.
›Enables jumping to a specific column from the go-to-line modal by typing :<column> after the line number.
Zed v0.85.3 adds ESLint diagnostics support and per-path Copilot suppression via glob patterns.
└──▷ GET THIS VERSION
$ git clone --branch v0.85.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:$ git checkout v0.85.3
›Adds 'Hide Suggestions for This Path' option in the Copilot menu to populate a glob-pattern setting that disables Copilot suggestions for specific file paths.
›Adds support for displaying ESLint diagnostics in JavaScript/TypeScript projects that have an ESLint configuration.
›Improves Markdown rendering in editor hover popovers.
$ git clone --branch 0.9.1 https://github.com/TheR1D/shell_gpt.git
# already have the repo? check out this version:$ git checkout 0.9.1
└──▷ USE IT
Auto-execute shell suggestions on Enter in CI or scripting workflows without typing 'e' each time.
ini
DEFAULT_EXECUTE_SHELL_CMD=true
Inspect a suggested command interactively before running it during a shell REPL session.
$ sgpt -s--repl temp
# >>> list running containers
# docker ps
# >>> d
# Lists all currently running Docker containers.
# >>> e
›Adds --describe-shell (or -d) CLI flag to generate a natural-language explanation of any shell command, e.g. sgpt -d "ls -la".
›Adds DEFAULT_EXECUTE_SHELL_CMD config parameter to .sgptrc (default false); set to true to auto-execute --shell suggestions on Enter without typing e.
›Adds [D]escribe option to the --shell execution prompt, letting users get an inline explanation of a suggested command before deciding to execute or abort.
›Adds [d] describe shortcut inside --repl shell mode, printing an explanation of the last suggested command before execution.
›Raises the maximum allowed value of --temperature to 2.
+1 moreshow less
›Excludes comment lines (lines starting with #) when parsing .sgptrc, enabling inline documentation in the config file.
›Adds built-in API key authentication support for securing Qdrant endpoints.
›Enables Product Quantization (PQ) for vectors, providing a new vector compression strategy alongside existing quantization options.
›Adds Group-By API, allowing search results to be grouped by a payload field for top-results-per-group retrieval.
›Adds optional vectors, allowing points to be created with only a subset of named vectors defined — two new APIs manage vectors independently from payload.
›Adds recovery mode for handling Out-of-Disk and Out-of-Memory errors, enabling the node to recover rather than crash.
Weaviate v1.19 adds a gRPC search API, Cohere generative module, tunable consistency, uuid prop types, and group-by queries.
└──▷ GET THIS VERSION
$ git clone --branch v1.19.0 https://github.com/weaviate/weaviate.git
# already have the repo? check out this version:$ git checkout v1.19.0
›Adds a minimal gRPC API (experimental) with support for a Search endpoint, enabling lower-latency programmatic access.
›Adds generative-cohere module, enabling Retrieval-Augmented Generation with Cohere's generative models.
›Adds tunable consistency to GraphQL Get queries, letting callers control read consistency level per search request.
›Adds uuid and uuid[] property types, indexed with roaring bitmaps for efficient UUID-based filtering.
›Adds group-by arbitrary property (including reference props) in GraphQL queries, returning top-k results per group.
+2 moreshow less
›Enriches text and text[] tokenization with new options via IndexFilterable and IndexSearchable property settings, replacing the deprecated string and string[] data types.
›Migrates the IndexInverted property field to separate IndexFilterable and IndexSearchable fields for finer control over inverted index behavior.
└──▷ BREAKING ON UPGRADE
!Downgrading from v1.19 to v1.18 is not supported after upgrading; a backup must be created before upgrading if a downgrade may be needed.
!The string and string[] data types are deprecated in favor of text and text[] with explicit tokenization options.
THE AI TOOLCHAIN