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.
Aider v0.7.0 adds --message for one-shot GPT instructions and --no-stream for token cost visibility.
└──▷ GET THIS VERSION
$ git clone --branch v0.7.0 https://github.com/Aider-AI/aider.git
# already have the repo? check out this version:$ git checkout v0.7.0
└──▷ TRY IT
Run a one-shot refactor instruction in CI without an interactive session.
$ aider --message 'Refactor all print statements to use logging' src/app.py
Disable streaming to capture token usage and cost info after a GPT edit session.
$ aider --no-stream src/app.py
›Adds --message flag to send a single instruction to GPT and exit after it replies and applies any edits — enabling non-interactive, scriptable use.
›Adds --no-stream flag to disable streaming GPT responses; non-streaming mode surfaces token usage info and displays cost estimates based on OpenAI advertised pricing.
›New coding competence benchmarking tool for evaluating GPT performance against a suite of programming tasks drawn from the Exercism Python repository.
›Initial function-based code editing backend for GPT-3.5, laying groundwork for OpenAI function calls API support.
›Adds graceful handling of context window exhaustion with user-facing tips when the limit is hit.
2 more releases in this issue
· 2023-06-07 → 2023-06-25
Aider v0.5.0 adds GPT-3.5/4-32k support, token budgeting with --map-tokens, and new /tokens and /clear chat commands.
└──▷ GET THIS VERSION
$ git clone --branch v0.5.0 https://github.com/Aider-AI/aider.git
# already have the repo? check out this version:$ git checkout v0.5.0
└──▷ TRY IT
Cap the repo map to a specific token budget so large codebases fit within context limits when using GPT-4-32k.
$ aider --map-tokens 4096 --model gpt-4-32k
Check how many tokens the current conversation is consuming before adding more context.
$ /tokens
›Adds --map-tokens CLI flag to set a token budget for the repo map, using a PageRank-based algorithm to prioritize which files and identifiers are included.
›Adds in-chat command /tokens to report current context window token usage.
›Adds in-chat command /clear to clear the conversation history.
›Adds support for gpt-3.5-turbo and gpt-4-32k models.
Zed v0.91.3 adds a syntax tree view, Elixir HEEx support, Ruby singleton methods in outline, and comment toggling for empty lines.
└──▷ GET THIS VERSION
$ git clone --branch v0.91.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:$ git checkout v0.91.3
└──▷ HOW TO FIND IT
Inspect the internal syntax tree of the file you are editing to debug grammar or extension issues.
📍debug: open syntax tree view
›Adds a syntax tree view showing Zed's internal parse tree for the active editor; open it via the debug: open syntax tree view command palette command.
›Adds support for Elixir HEEx templates.
›Adds support for singleton methods in the Ruby outline view.
›Adds the ability to toggle comments on empty single-line selections.
shell-gpt 0.9.2 adds ZSH and Bash shell integration with hotkey-triggered command suggestions via --install-integration.
└──▷ GET THIS VERSION
$ git clone --branch 0.9.2 https://github.com/TheR1D/shell_gpt.git
# already have the repo? check out this version:$ git checkout 0.9.2
└──▷ TRY IT
Install the shell integration once so that Ctrl+l triggers GPT command suggestions from any terminal prompt.
$ pip install --upgrade shell-gpt
sgpt --install-integration
# Restart your terminal, then type a natural-language request and press Ctrl+l to get a command suggestion.
›Adds --install-integration hidden flag to automatically update the user's shell profile and enable shell integration for ZSH and Bash.
›New shell integration for ZSH and Bash lets users press Ctrl+l in the terminal to replace the current input buffer with an sgpt-suggested shell command.
›Shell integration surfaces sgpt completions inside shell history and allows immediate in-line editing of suggested commands.
$ git clone --branch v0.4.3 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout v0.4.3
›Adds replace_in_file command to modify files by replacing occurrences of text without loading the entire file into memory.
›Adds support for OpenAI Functions, laying groundwork for function-calling-based command execution.
›Adds run_task Python hook to interface AutoGPT with benchmarks, with all challenges linked to it.
›Extends supported OpenAI model info to include the GPT-3 16k model.
›Improves JSONFileMemory to support loading MemoryItems from file, enabling persistent memory across sessions.
+4 moreshow less
›Switches token counting to use tiktoken for more precise cost estimates.
›Prompts users for their OpenAI API key interactively if not already set in the environment.
›Automatically unregisters commands incompatible with the current configuration.
›Restricts plugin loading to subclasses of AutoGPTPluginTemplate, tightening plugin validation.
└──▷ BREAKING ON UPGRADE
!The commands analyze_code, write_tests, improve_code, audio_text, web_playwright, and web_requests have been removed from the core application and are no longer available without a plugin.
2 more releases in this issue
· 2023-06-05 → 2023-06-28
AutoGPT v0.4.1 adds directory-based plugins, new replace_in_file and Python execution commands, keyboard navigation, and per-model token limits.
└──▷ GET THIS VERSION
$ git clone --branch v0.4.1 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout v0.4.1
›Adds replace_in_file command to change occurrences of text in a file without rewriting the whole file.
›Adds a command for directly executing Python code, including within a Docker install.
›Implements directory-based plugin system — plugins can now be loaded from a subfolder or repository clone inside the plugins directory, no zip file required.
›Reworks plugin config to be file-based.
›Adds settings for a custom base URL for the OpenAI API endpoint.
+5 moreshow less
›Sets token limits automatically based on the model in use.
›Enables keyboard navigation in the CLI via prompt_toolkit.
›Implements batch summarization in the MessageHistory class to keep context length within the model's token limit.
›Streamlines shell command control configuration (allowed/denied commands) to reduce ambiguity.
›Automatically retries DuckDuckGo searches when rate limits are detected, improving search reliability.
AutoGPT v0.4.0 adds document reading in read_file, granular command-access controls via DISABLED_COMMAND_CATEGORIES, DENY_COMMANDS, and ALLOW_COMMANDS, and revamps the vector memory system.
└──▷ GET THIS VERSION
$ git clone --branch v0.4.0 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout v0.4.0
└──▷ USE IT
Restrict AutoGPT to only safe, pre-approved shell commands in a production or sandboxed environment.
Haystack v1.18 adds AWS SageMaker LLM support, PromptHub integration, ConversationalAgent tools, and a new CohereRanker node.
└──▷ GET THIS VERSION
$ git clone --branch v1.18.0 https://github.com/deepset-ai/haystack.git
# already have the repo? check out this version:$ git checkout v1.18.0
└──▷ USE IT
Pull a ready-made prompt from PromptHub by name to classify topics without writing a prompt from scratch.
python
import os
from haystack.nodes import PromptNode, PromptTemplate
template = PromptTemplate("deepset/topic-classification")
prompt_node = PromptNode(
model_name_or_path="text-davinci-003",
api_key=os.environ.get("OPENAI_API_KEY")
)
result = prompt_node.prompt(
prompt_template=template,
documents="YOUR_DOCUMENTS",
options=["sports", "politics", "technology"]
)
Equip a ConversationalAgent with a QA pipeline tool so it can answer domain-specific questions mid-conversation.
python
from haystack.agents import Tool
from haystack.agents.conversational import ConversationalAgent
search_tool = Tool(
name="USA_Presidents_QA",
pipeline_or_node=presidents_qa_pipeline,
description="useful for when you need to answer questions about US presidents."
)
agent = ConversationalAgent(prompt_node=prompt_node, tools=[search_tool])
agent.run("Who was the 35th president of the United States?")
›Adds AWS SageMaker-hosted LLM support to PromptNode via model_kwargs keys aws_profile_name and aws_region_name, enabling open-source models deployed on SageMaker endpoints.
›Introduces PromptHub integration: PromptTemplate now accepts a hub prompt name (e.g. 'deepset/topic-classification') directly, with local caching of fetched prompts.
›Adds tools parameter to ConversationalAgent for attaching Tool instances (pipelines or nodes) to a chat agent.
›Adds prompt_template parameter to ConversationalAgent.__init__ for customising the agent's prompt at construction time.
›Adds CohereRanker node backed by the Cohere reranking endpoint.
+8 moreshow less
›Adds batch_size parameter to WeaviateDocumentStore query methods.
›Adds batching support for querying in ElasticsearchDocumentStore and OpenSearchDocumentStore.
›Adds current_datetime shaper function for use in pipeline prompt construction.
›Adds max_chars_check hard document length limit to pipeline processing.
›Adds optional content moderation for OpenAIPromptNode and OpenAIAnswerGenerator.
›Supports passing model parameters to HFLocalInvocationLayer via model_kwargs for direct model usage.
›Supports setting a custom api_base for OpenAI nodes.
›New farm-haystack[inference] extra installs PyTorch and related dependencies for local model execution, keeping the base install lighter for API-only users.
└──▷ BREAKING ON UPGRADE
!PromptTemplate no longer accepts name or prompt_text parameters; use prompt and output_parser instead.
!Seq2SeqGenerator and RAGenerator have been removed; use PromptNode instead.
!The deprecated PDFToTextOCRConverter node has been removed.
!The deprecated return_table_cell parameter has been removed.
!PyTorch and inference-related dependencies are no longer installed by default; run pip install farm-haystack[inference] to restore local model support.
!Weaviate authentication has been simplified (feat!: simplify weaviate auth); existing auth configuration may need to be updated.
LangChain v0.0.218 adds MultiQueryRetriever, new document loaders, OAuth for Zapier, proxy support for WebBaseLoader, and async Zapier NLA tools.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.218 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.218
└──▷ USE IT
Use MultiQueryRetriever to improve recall by automatically generating multiple query phrasings from a single question.
python
from langchain.retrievers.multi_query import MultiQueryRetriever
retriever = MultiQueryRetriever.from_llm(
retriever=vectorstore.as_retriever(),
llm=llm
)
docs = retriever.get_relevant_documents(query="What are the security implications of prompt injection?")
›Adds UnstructuredOrgModeLoader for loading Org-mode documents.
›Adds MultiQueryRetriever to generate multiple query variations and merge results for improved retrieval coverage.
›Adds source code loader based on AST manipulation for structured code document loading.
›Adds Tencent COS directory and file document loaders.
›Adds LarkSuite document loader.
+7 moreshow less
›Adds proxy support to WebBaseLoader.
›Adds optional HTTP error exception raising to WebBaseLoader.
›Adds async support to Zapier NLA tools.
›Adds OAuth support to the Zapier integration.
›Adds streaming of only the final output via async iteration for agents.
›Allows rail_parser to be created from Pydantic models.
›Enhances WhatsAppChatLoader to ignore deleted messages and media.
$ git clone --branch v0.0.210 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.210
└──▷ USE IT
Stream an agent's reasoning steps live into a Streamlit app for real-time visibility during a run.
python
import streamlit as st
from langchain.callbacks import StreamlitCallbackHandler
from langchain.agents import initialize_agent, AgentType
from langchain.llms import OpenAI
llm = OpenAI(streaming=True)
agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION)
with st.container():
handler = StreamlitCallbackHandler(st.container())
agent.run("What is the weather in San Francisco?", callbacks=[handler])
Purge specific entries from a Redis-backed memory or cache by key to keep it clean between sessions.
python
from langchain.vectorstores.redis import Redis
redis_store = Redis.from_existing_index(embedding=embeddings, index_name="my-index")
redis_store.delete(["doc:abc123", "doc:def456"])
Tag agent runs at initialization so you can filter them by environment or experiment in your tracing project.
LangChain v0.0.190 adds UnstructuredExcelLoader, PubMed integration, FileCallbackHandler, PipelinePrompt, and Personal Access Token auth for Confluence.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.190 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.190
└──▷ USE IT
Load an Excel spreadsheet into LangChain documents for indexing or QA.
python
from langchain.document_loaders import UnstructuredExcelLoader
loader = UnstructuredExcelLoader('report.xlsx')
docs = loader.load()
›Adds UnstructuredExcelLoader class for loading .xlsx and .xls files as documents.
›Adds FileCallbackHandler for writing chain and agent callback events to a file.
›Adds Personal Access Token authentication support to ConfluenceLoader.
›Adds similarity_score_threshold retrieval mode support to Chroma vector store.
›Adds PubMed integration as a new data loader/tool.
+4 moreshow less
›Adds pipeline prompt support (PipelinePromptTemplate) for composing prompts from sub-prompts.
›Adds the option to pass the original prompt into AgentExecutor for PlanAndExecute agents.
›Adds MongoDBChatMessageHistory index creation on SessionId for improved query performance.
›VertexAI chat models (PaLM2) now accept additional parameters on send_message() calls.
└──▷ BREAKING ON UPGRADE
!Weaviate integration removes client and namespace configuration in favor of collection.
›Adds --usecublas flag to select CUDA inference mode, with --usecublas lowvram for low-VRAM operation, alongside --useclblast for CLBlast — all available in a single All-In-One distributable.
›Adds --contextsize flag to enable long-context inference beyond 2048 tokens via Scaled RoPE for LLAMA models (tested up to 3k+).
›Extends CUDA GPU layer offloading to MPT, GPT-2, GPT-J, and GPT-NeoX model architectures.
›Back-ports CUDA support to all prior GGML file format versions (GGML, GGMF, GGJT v1/v2/v3), allowing legacy quants to load and run under CUDA mode.
›Applies VRAM memory optimizations (previously OpenCL-only) to CUDA, enabling more layers to be offloaded than upstream llama.cpp.
+1 moreshow less
›Instruct mode now supports any number of newlines in start and end tags, configurable by the user.
6 more releases in this issue
· 2023-06-01 → 2023-06-29
KoboldCpp v1.32.3 brings OpenCL K-Quant CUDA kernel ports and GPU offloading for MPT, GPT-2, GPT-J, and GPT-NeoX architectures.
└──▷ GET THIS VERSION
$ git clone --branch v1.32.3 https://github.com/LostRuins/koboldcpp.git
# already have the repo? check out this version:$ git checkout v1.32.3
›Ports optimized K-Quant CUDA kernels to OpenCL, delivering approximately 15% faster generation speed for K-Quant models on OpenCL backends.
›Adds basic GPU offloading via OpenCL for MPT, GPT-2, GPT-J, and GPT-NeoX architectures, achieving roughly 50% speedup for GPT-J and 30% for WizardCoder (weights still copied to RAM).
›Implements scratch buffers for MPT, GPT-2, NeoX, and GPT-J architectures, significantly reducing BLAS memory usage and enabling larger BLAS batch sizes on these models.
›Extends GPT tokenizer with support for Starcoder special added tokens, improving coherence for non-llama models.
›Adds --hordeconfig support for specifying a separate max context length allowed in horde, independent of the real context length used for memory allocation.
›Displays token generation rate in the console during inference.
›Gates server endpoint debug output behind --debugmode; also suppresses incoming horde prompt logs when --hordeconfig is set unless --debugmode is also enabled.
›Adds new scenarios, italics chat text, and customizable idle message counts to Kobold Lite.
└──▷ BREAKING ON UPGRADE
!RWKV sequence mode is disabled; setups relying on it for speed improvements will no longer benefit from those gains.
›Adds a true SSE-streaming endpoint at /api/extra/generate/stream (Agnaistic-compatible) that streams tokens in real time during generation.
›Adds a polled-streaming endpoint at /api/extra/generate/check to fetch in-progress results without disrupting generation; the old 8-token chunked streaming remains available via streamamount=8 in the URL.
›Adds an abort endpoint at /api/extra/abort that halts in-progress generation without stopping the server, integrated into Kobold Lite via an 'abort' button.
›Adds --lora [lora_file] [base_model] optional second parameter for specifying a LoRA base model.
›Adds full OpenCL / CLBlast support for K-Quants, covering both prompt processing and GPU offloading across all K-quant formats.
+2 moreshow less
›Adds support for the RWKV World Tokenizer and RWKV-World models, with automatic detection and selection.
›Adds CUDA support for older ggjtv2 models via backport, available through koboldcpp_CUDA_only.exe.
›Renames --renamemodel to --hordeconfig, which now accepts two parameters: the horde display name and the advertised max generation length on horde.
›Adds --blasbatchsize -1 to disable BLAS while retaining CLBlast GPU layer offloading, enabling more layers to be offloaded and faster generation at the cost of slower prompt processing.
›Adds support for MPT models; use --contextsize to set context length (up to ~5000 tokens tested).
›Enhances the KoboldCpp Easy Launcher GUI with controls for the number of CLBlast GPU layers, thread count, and additional toggles.
›Improves CLBlast memory allocation efficiency, allowing more layers to be offloaded to GPU than before.
└──▷ BREAKING ON UPGRADE
!The --renamemodel flag has been renamed to --hordeconfig; any scripts or configs using --renamemodel will break on upgrade.
KoboldCpp v1.27 adds --renamemodel, RWKV v101 support, AutoGenerated Memory in Kobold Lite, and CL BLAST VRAM-only layer offloading.
└──▷ GET THIS VERSION
$ git clone --branch v1.27 https://github.com/LostRuins/koboldcpp.git
# already have the repo? check out this version:$ git checkout v1.27
└──▷ TRY IT
Override the model name reported to the Horde network so it appears under a custom identifier.
$ koboldcpp.exe --renamemodel mymodel
›Adds --renamemodel CLI flag to override the displayed model name (prefixed with koboldcpp/), useful for Horde operators.
›Supports RWKV.cpp v101 model format (updated GGML library, smaller and faster), with automatic backwards-compatible detection of both old and new quantization formats.
›Adds EOS token support for RWKV models.
›Integrates CLBlast GPU offloading improvement that stores layers fully in VRAM without a duplicate RAM copy, reducing overall RAM usage when offloading GPU layers.
›Adds AutoGenerated Memory feature in Kobold Lite, which generates a short story summary/memory with a single click (works best with instruct models).
›Adds Stream: true support to the completion endpoint, enabling server-sent event streaming for inference responses.
›Adds LangChainGo HuggingFace backend, allowing LocalAI to route completions through HuggingFace-hosted models via LangChainGo.
›Adds experimental support for falcon-7b models.
›Adds upscaled image generation using esrgan.
›Adds Metal GPU support for gpt4all on Apple Silicon.
+5 moreshow less
›Updates llama.cpp to support k-quants model quantization formats.
›Enables Stable Diffusion by default in container images.
›Adds ffmpeg-enabled Docker image variants for audio processing use cases.
›Enables static builds for Linux binaries, removing dynamic library dependencies.
›Builds multiple Docker image variants via CI, expanding deployment options.
└──▷ BREAKING ON UPGRADE
!gpt4all is updated with support for multiple implementations selectable at runtime; existing gpt4all configurations may need to be updated to specify the desired implementation.
Triton v2.35.0 adds Redis cache support, gRPC response flags, KIND_MODEL for PyTorch, and serial sequences in Perf Analyzer.
└──▷ GET THIS VERSION
$ git clone --branch v2.35.0 https://github.com/triton-inference-server/server.git
# already have the repo? check out this version:$ git checkout v2.35.0
└──▷ TRY IT
Run Perf Analyzer in serial sequences mode to avoid interleaving requests across sequences during load testing.
$ perf_analyzer -m my_model --serial-sequences
›Adds --serial-sequences mode to Perf Analyzer for sequential load testing of inference requests.
›Adds beta support for using Redis as a cache for inference requests via the redis_cache backend.
›Extends KIND_MODEL instance group type support to the PyTorch backend.
›The statistics extension now includes memory usage of loaded models (currently implemented for TensorRT and ONNXRuntime backends).
›gRPC clients can now request response flags per response to programmatically determine when all responses for a decoupled model request have been received.
+1 moreshow less
›Adds support for batch inputs in ragged batching for the PyTorch backend.
Phoenix v0.0.24 adds cluster-level performance and data quality metrics including accuracy_score and ad-hoc numeric averages via GraphQL.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.24 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout v0.0.24
›Adds a standalone clusters endpoint to the GraphQL API, enabling programmatic retrieval of cluster data outside the embedding projection UI.
›Adds prediction_id to GraphQL event metadata and surfaces it in selection details, making it easier to trace individual predictions back to source records.
›Adds accuracy_score as a model performance metric on embedding clusters, letting practitioners sort clusters by worst accuracy to surface pockets of bad predictions.
›Adds cluster-level data quality / custom metrics: calculates the average of any numeric feature, tag, prediction, or actual value (e.g., precision@k, rouge score, LLM-assisted eval scores) per cluster.
›Adds a metric selector UI for embeddings, allowing users to switch between performance and data quality metrics on the cluster view.
+3 moreshow less
›Adds cluster sorting by metric value in the embeddings UI, so the worst-performing or lowest-quality clusters rise to the top.
›Adds accuracy timeseries tracking for clusters, extending the performance metric beyond point-in-time analysis.
›Adds data_type filter support on dimensions, enabling finer control over which features are included in analysis.
›Adds consistency_level field to search/query requests, letting callers control read consistency per operation.
›Introduces a native JSON data type for collection fields, with expression support and bulk-insert compatibility for JSON data.
›Adds dynamic schema support, allowing fields to be added to a collection without a fixed schema definition, including in bulk insert operations.
›Introduces partition keys to route data within a single collection across logical partitions — enabling multi-tenant data separation without separate collections.
›Extends RBAC to the database level, scoping roles and privileges to specific databases for stronger multi-tenancy control.
+7 moreshow less
›Implements a connect API for explicit connection management, including existence checks for the target database at connect time.
›Adds Alibaba Cloud OSS as a supported object storage backend, configurable via access key (AK) or IAM authentication.
›Implements AutoIndex to automatically select index parameters without requiring manual tuning.
›Adds a configurable scheduling policy for query nodes, with user-level schedule policy support.
›Implements rate limiting based on growing segment size to protect write throughput.
›Adds SIMD-accelerated evaluation for several filtering expressions, improving filtered-search performance.
›Adds support for single-quoted string literals inside filter expressions.
└──▷ BREAKING ON UPGRADE
!Milvus standalone deployments using MinIO standalone have incompatible data layouts; manual data migration to a new MinIO instance is required before upgrading (see https://min.io/docs/minio/linux/operations/install-deploy-manage/migrate-fs-gateway.html).
Qdrant v1.3.0 adds a self-hosted Web UI, group lookup, io_uring async IO, and quantized vector oversampling.
└──▷ GET THIS VERSION
$ git clone --branch v1.3.0 https://github.com/qdrant/qdrant.git
# already have the repo? check out this version:$ git checkout v1.3.0
└──▷ HOW TO FIND IT
Explore collections and run ad-hoc vector queries with autocomplete against your local Qdrant instance — no client library needed.
📍Open http://localhost:6333/dashboard in your browser, select a collection, and use the interactive query editor to run searches.
›Adds experimental io_uring async IO support (opt-in) for significantly improved performance on network-mounted storages, using a recent Linux kernel feature.
›Adds oversampling for quantized vector queries, letting you retrieve more candidate points with quantized vectors and re-score with originals to tune the speed/accuracy tradeoff at query time.
›Adds lookup in grouping requests, enabling shared group metadata to be stored in a dedicated collection to minimize memory usage.
›Adds a self-hosted Web UI dashboard at http://localhost:6333/dashboard, including a collection viewer and an interactive query editor with autocomplete.
THE AI TOOLCHAIN