Hugging Face Transformers: Patch release v5.22.1
Hugging Face Transformers v5.22.1 is a minor documentation-only patch release. It fixes an issue by removing a work-in-progress document for Ml4.
AI Models & LLM Frameworks archive — Page 1
1172 tracked entries. Older history is imported progressively; see source coverage.
Hugging Face Transformers: v5.22.0: Mistral Large 4 (Le Chonk)
Hugging Face Transformers v5.22.0 introduces the Mistral Large 4 1T-parameter model and adds support for FP8 adapter training. The release includes a breaking architectural change regarding Execution Provider (EP) logic separation.
Open WebUI v0.12.5 — Release Notes & Engineering Impact
Open WebUI v0.12.5 introduces an Analytics session explorer, multi-page notes, and live presence alongside major performance optimizations for large knowledge bases and chat searches. It also includes critical security fixes and resolves token counting calculation inflations.
Hugging Face Transformers: Patch release v5.21.1
Hugging Face Transformers v5.21.1 is a patch release that fixes the rope_scaling setter for configurations that assign it during initialization. This update resolves a critical regression that broke remote code loading in vLLM.
Ollama v0.40.13 — Release Notes & Engineering Impact
Version 0.40.13 expands support for Liquid AI’s LFM3 models—including text‑to‑speech variants lfm3‑audio and lfm3‑omni—to Linux and Windows, enabling cross‑platform usage beyond Apple silicon.
Ollama v0.40.12 — Release Notes & Engineering Impact
v0.40.12 adds the LFM3 family of small multimodal models and d1 2B decision models, expanding local AI capabilities on Apple silicon. It also delivers performance improvements and bug fixes without introducing breaking changes.
Hugging Face Transformers: Release: v5.21.0
Hugging Face Transformers v5.21 introduces support for multiple new models (such as MiniCPM-OCR, Flint, UltraVBert, and Gradium/Pocket TTS) and enforces breaking architectural changes including custom tensor parallelism and MoE expert sharding. Developers upgrading to this version should review updated TP plans and custom MoE implementations.
Open WebUI v0.12.4 — Release Notes & Engineering Impact
Open WebUI v0.12.4 introduces major performance enhancements such as skill lazy loading and context compaction alongside granular permission enforcement and extensive UI accessibility upgrades. Users will experience a more streamlined interface with context-aware controls, improved state persistence, and critical security fixes.
Ollama v0.40.11 — Release Notes & Engineering Impact
Ollama v0.40.11 adds speech generation on macOS via the `ollama run` command and an OpenAI-compatible `/v1/audio/speech` endpoint, plus introduces d3 decision models for classification, yes/no answers, and probabilistic scoring. These features expand Ollama's multimodal capabilities and structured reasoning without introducing breaking changes.
Ollama v0.40.10 — Release Notes & Engineering Impact
Ollama version v0.40.10 introduces support for NVIDIA's Nemotron 3.5 Bolt E4B vision-language model. Developers can now run this 128K-token context model locally with speculative decoding support.
LocalAI v4.12.1 — Release Notes & Engineering Impact
LocalAI v4.12.1. Source highlight: Patch release v4.12.1. Security fix only.
Dify: v1.18.0 — Smarter Agent Routing, Portable Apps, Deeper Workflow Visibility
Dify 1.18.0 introduces smart agent workflow routing, portable `.ifpkg` app packages, nested workflow trace visibility, and an opt-in Agent Knowledge Base (KnowledgeFS). Upgrading requires simultaneous component updates, Python 3.13 for source installs, a mandatory Local sandbox authentication token, and careful handling of database migrations.
LocalAI v4.12.0 — Release Notes & Engineering Impact
LocalAI v4.12.0 introduces a redesigned web UI, streaming voice responses, native decision routing, and persistent Qdrant vector stores. The release also implements advanced privacy controls, remote MCP OAuth2 authentication, and live GEM-X motion capture.
Ollama v0.40.9 — Release Notes & Engineering Impact
Ollama v0.40.9 brings Flint support to Linux and Windows platforms while improving error handling for streaming OpenAI and Anthropic API endpoints. Additionally, it resolves context length propagation bugs in `ollama launch` and refines thinking configuration logic for web search follow-up requests.
LangChain langchain==1.4.7 — Release Notes & Engineering Impact
LangChain langchain==1.4.7. Source highlight: Changes since langchain==1.4.6
Ollama v0.40.8 — Release Notes & Engineering Impact
Ollama v0.40.8. Source highlight: Flint brings Recursive's coding model to Ollama. It is designed for the work coding agents do across a repository: understanding unfamiliar code, tracing bugs, implementing features, and adding tests. Reasoning, multi-turn tool use, and a 262K-token context window support longer tasks across files a
Cursor: Agent SDK
Cursor agent-sdk. Source highlight: The Agent SDK is available today. It allows you to define an agent as a folder of code, test it with evals, and deploy it to Cursor or your own infrastructure.
Hugging Face Transformers: Patch release v5.20.2
Hugging Face Transformers v5.20.2 is a patch release addressing critical bugs in multimodal generation inputs, bfloat16 expert parallelism gradients, and eager_mask host-to-device copies. It ensures correct input typing within `prepare_inputs_for_generation` and prevents performance regressions in masking.
vLLM v0.32.0 — Release Notes & Engineering Impact
vLLM v0.32.0 introduces Model Runner V2 as the default, integrates comprehensive structured-output support, and enforces significant dependency and configuration breaking changes. Users must prepare for Python 3.11 requirements, updated CORS defaults, and API shifts in reasoning streaming.
Ollama v0.40.7 — Release Notes & Engineering Impact
Ollama v0.40.7. Source highlight: pplx-decider is a 27B decision model from Perplexity, fine-tuned from
Ollama v0.40.6 — Release Notes & Engineering Impact
Ollama v0.40.6. Source highlight: win: prevent orphaned runners
Hugging Face Transformers: Release v5.20.1
Hugging Face Transformers v5.20.1. Source highlight: Special surprise including the new model Beam from Reflection AI :partying_face:
Open WebUI v0.12.3 — Release Notes & Engineering Impact
Open WebUI v0.12.3 introduces end-to-end encrypted chats, optimized Redis-based streaming, and robust workspace search capabilities. It also resolves critical authentication, OAuth token refresh, and audit logging issues.
Cursor: Cloud Agent Canvases and Environment Improvements
Cursor cloud-agent-canvases-and-environ. Source highlight: Cloud Agent Canvases and Environment Improvements
Hugging Face Transformers: Release v5.20.0
Hugging Face Transformers v5.20.0. Source highlight: Muse Spark Open is a multimodal mixture-of-experts model from Meta that takes text, images, audio, and video as input and generates text. It has 816B total parameters (42B active per token) and a context window of up to 1,048,576 tokens shared across modalities. It is trained for agentic coding task
Open WebUI v0.12.2 — Release Notes & Engineering Impact
Open WebUI v0.12.2 introduces support for self-hosted MinerU 4.x document servers, custom TLS configurations for Milvus, and enhanced model ID tracking in headers and audit logs. The release also includes critical security hardening for SCIM user deletion and updates Valkey vector store connections to require valkey-glide-sync 2.5.2 or newer.
Ollama v0.40.5 — Release Notes & Engineering Impact
Ollama v0.40.5. Source highlight: NVIDIA Nemotron 3.5 Super
LangChain langchain==1.4.6 — Release Notes & Engineering Impact
LangChain langchain==1.4.6. Source highlight: Changes since langchain==1.4.5
Open WebUI v0.12.1 — Release Notes & Engineering Impact
Open WebUI v0.12.1 introduces major new architecture support including external channel apps, Azure OpenAI Realtime calls, and a comprehensive skill archiving/sharing system. The release contains breaking database migrations requiring synchronized updates across multi-worker environments and a mandatory backup before upgrading.
Cursor: Origin App Marketplace
Cursor origin-app-marketplace. Source highlight: Origin App Marketplace
LangChain langchain==1.4.5 — Release Notes & Engineering Impact
LangChain langchain==1.4.5. Source highlight: Changes since langchain==1.4.4
Ollama v0.40.4 — Release Notes & Engineering Impact
Ollama v0.40.4. Source highlight: Decision models in the CLI
Ollama v0.40.3 — Release Notes & Engineering Impact
Ollama v0.40.3. Source highlight: `ollama launch claude` no longer shows a warning about Claude connectors
Open WebUI v0.12.0 — Release Notes & Engineering Impact
Open WebUI v0.12.0. Source highlight: ☎️ **Realtime voice calls.** Administrators can now switch Call mode in the Voice calls section of Audio settings from Standard to Realtime, so Voice mode talks through an OpenAI Realtime voice model (gpt-realtime-2.1-mini with the Marin voice by default) that handles small talk itself and hands rea
LangChain langchain==1.4.4 — Release Notes & Engineering Impact
LangChain langchain==1.4.4. Source highlight: Changes since langchain==1.4.3
Ollama v0.40.2 — Release Notes & Engineering Impact
Ollama v0.40.2. Source highlight: Models downloaded with earlier versions of Ollama are upgraded in the background the first time you run them, for better performance and compatibility when running on llama.cpp.
Ollama v0.40.1 — Release Notes & Engineering Impact
Ollama v0.40.1. Source highlight: server: proxy cloud usage and balance APIs
Hugging Face Transformers: Release v5.19.0
Hugging Face Transformers v5.19.0. Source highlight: EmbeddingGemma 2 is a multimodal embedding model from Google built on the Gemma 4 architecture. It encodes text, images, audio, and video, individually or combined in one input, into a shared 768-dimensional vector space for cross-modal retrieval, semantic similarity, clustering, and classification.
Cursor: Remote control for local agents
Cursor remote-control-local-agents. Source highlight: Remote control for local agents
vLLM v0.31.0 — Release Notes & Engineering Impact
vLLM v0.31.0 introduces major architectural performance additions for DeepSeek-V4.1-Flash alongside strict security gating for multimodal request arguments and configuration parameter renames. Upgrading teams must review breaking changes around runtime parameters, removed backends, and modified quantization APIs.
LocalAI v4.11.0 — Release Notes & Engineering Impact
LocalAI v4.11.0. Source highlight: 🎉 LocalAI 4.11.0 Release! 🚀
Hugging Face Transformers: Release 5.18.0
Hugging Face Transformers v5.18.0. Source highlight: Nemotron 3 Diarization
Ollama v0.35.1 — Release Notes & Engineering Impact
Ollama v0.35.1. Source highlight: Clef decision models
Ollama v0.35.0 — Release Notes & Engineering Impact
Ollama v0.35.0. Source highlight: Ollama now supports decision models through `/v1/systemone`, based on TypeSafe’s Jev API.
LangChain langchain==1.4.3 — Release Notes & Engineering Impact
LangChain langchain==1.4.3. Source highlight: Changes since langchain==1.4.2
Ollama v0.40.0 — Release Notes & Engineering Impact
Ollama v0.40.0. Source highlight: **Models run on MLX on Apple Silicon by default**
Ollama v0.34.4 — Release Notes & Engineering Impact
Ollama v0.34.4. Source highlight: Structured outputs on thinking models now apply in a single pass, making them faster and more reliable.
Cursor: Rollouts and Security Review
Cursor rollouts-and-security-reviewer. Source highlight: Rollouts and Security Review
vLLM v0.30.0 — Release Notes & Engineering Impact
vLLM v0.30.0. Source highlight: This release features 762 commits from 315 contributors (104 new)!
Open WebUI v0.11.4 — Release Notes & Engineering Impact
Open WebUI v0.11.4. Source highlight: 📉 **Far smaller slim image.** A slim build now comes down at around 175 MB, near enough 89% smaller than the last release, the local models, the packages around them and the tools that installed them all gone from it; what that changes about the way an instance behaves is set out under Changed below