Cada lançamento que importa, lado a lado: novas versões das melhores ferramentas de IA open-source — modelos, executores, agentes, geradores de imagem — e cada novo modelo que chega nas APIs comerciais. Atualizado diariamente.
Carregando…
What's Changed 🚀 Features * Add a first-run banner describing the run, and open clai sessions with it by @DouweM in https://github.com/pydantic/pydantic-ai/pull
20.11.0 (2026-09-12) · Features · * annotations: add GraphQL mutations for span, trace, and session notes (#16053) (e374a4d) * mcp: serve the shared skills root
1.811.0 (2026-09-12) · Features · * make snowflake_oauth work as a dbt warehouse on every engine (#11095) (9fc50a2) · Bug Fixes · * accept any hub version of th
What's Changed * feat(artifacts): add first-class interactive HTML artifacts by @AnishSarkar22 in https://github.com/MODSetter/SurfSense/pull/1729 * feat(artifa
This release focuses on significantly improving performance on large scale scenes for both CPU and GPU. The speed is now slowing down sub-linearly wrt the numbe
Signed and notarized PuppyOne Desktop release. · Commit: 98f296043c034b6671db9c355f0d111957b2a3d4 Promoted Internal release: v0.3.19-internal.39 R2 version pref
📦 Release v2.2.17 · This release was automatically published from PR #19434. · Changes See PR description: https://github.com/lobehub/lobehub/pull/19434 · Commi
The composer now shows the current branch's GitHub pull request — PR number, merge status, changed-line totals, and CI checks. Click through to open it in your
Changes since langchain-core==1.6.2 · release(core): 1.6.3 (#40407) feat(core): Allow model name and provider tracing metadata override based on gateway respons
LoopX 1.0.3 — reliable manager delegation, atomic Todo planning, and wider host support · At a Glance · LoopX 1.0.3 makes the operator-to-Agent loop more depend
2.38.7 (2026-09-11) · Bug Fixes · * core: Execute MCP toolkit members on workers (#38277) (1025dc7) * core: Reuse an existing consent grant after the visitor au
Breaking Changes *none* · New Features *none* · Fixes * Repeating startup banner by @g-despot in https://github.com/weaviate/weaviate/pull/12912 * chore(dynamic
1.18.0 (2026-09-11) · What's Changed Features * feat: click an image attachment to open it full size by @palrohitg in https://github.com/OpenHands/OpenHands/pul
What's Changed · - chore: change aphrodite ascii art to sonar (#1785) (82b0cd8c2) by @lucyknada · Full Changelog: https://github.com/dphnAI/sonar/compare/v0.24.
OpenClaw v2026.9.4 · 20 direct commits · 1,558 pull requests · 294 contributors · Changes included in this release · The release notes and changelog contain the
Hermes Agent v0.21.2 (v2026.9.11) — The state.db Patch Release · Release Date: September 11, 2026 · > Patch release. v0.21.0 shipped a large rewrite of the sess
Overview · New version has been released. · Nightly build: b5130 More info: dist : releases and versioning of ggml-org projects · Changelog since v1.9.3 · 927cf
MCP · - Serve MCP Apps: AG2 can now serve MCP Apps, with structured tool output and extension support. - Input requests, both directions: An AG2 MCP server can
What's Changed · Features * feat(filters): unify sidebar search and saved views by @nkabardin in https://github.com/langfuse/langfuse/pull/17316 * feat(experime
Changelog · 🚀 Features ai - a3ab8b5 surface the AI Copilot with a button in the top bar (#19393), closes #19393 · core - 6c75b62 report translation keys that re
New · - Run dequeuing is now significantly faster when concurrency pools are in use and many runs are queued, as pool state is read once per dequeue pass rather
We're happy to announce the 1.9.1 release. · This release contains a few bug fixes and is the first version supporting Python 3.15. · You can see the changelog
> [!WARNING] > Self-hosted deployments using the bundled Weaviate must complete a manual, staged upgrade before starting 1.17.1. The bundled Weaviate server mov
Axolotl v0.19.0 Release Notes · 72 commits since v0.18.0 (July 17, 2026). · We have a variety of changes this release including a new declarative model-support
New features - Agentic RAG: Refactored the Agentic RAG retrieval framework, significantly improving reasoning speed and benchmark performance. (#19046, #19112,
v1.23.2 - 2026.09.10 · Downloads · - macOS (Apple Silicon) - macOS (Intel) - Windows (x64) - Linux (x64 DEB) - Linux (ARM64 DEB) - Android · What’s Changed · 1.
Features · Training beyond 1M tokens · · A new long-context guide plus a runnable example that trains a book-length sequence per step on a single 8×H100 node. ·
Changelog · 2.9.0 (2026-09-10) · Highlights · This release focuses on agent resilience and ecosystem integration by introducing automatic model failover, native
Verify Docker Image Signature · All LiteLLM Docker images are signed with cosign. Every release is signed with the same key introduced in commit 0112e53. · Veri
0.123.0 (2026-09-10) · ⚠ BREAKING CHANGES · * providers: default GPT-5.6+ to Responses (#10803) · Features · * assertions: grade native audio with llm-rubric (#
Added · - \[Server API\] Documentation for organization filtering parameters (org, org_id, and the X-Organization header) for endpoints that return object lists
What's Changed * [Partner Nodes] chore(Google): drop retiring Veo 2 and Veo 3.0 models by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/15883 * [Partne
v0.29.0 · Highlights · This release features 594 commits from 277 contributors (91 new)! · * Model Runner V2 is now the default for all models (#53183), complet
What's Changed · Features - Add telemetry to track checkpoint runtime and CLI usage. · Bug Fixes - Fix gateway errors reported inside an HTTP 200 response. - Ke
v3.0.1 · Release date: September 9, 2026 · | Milvus Version | Python SDK Version | Node.js SDK Version | Java SDK Version | Go SDK Version | | -------------- |
This is a large performance and reliability + bug fix release for Unsloth · Highlights * 1.2-1.7x faster diffusion. AMD 20% perf boost vs ROCM via Vulkan * 2x f
What's Changed · Version bumped from 1.0.52 to 1.0.53 · Recent Changes - Bump version to 1.0.53 (#8484) (a2a928b39) - Record keypress gestures in Record Browser
Core Improvements - Added the Astra system prompt for GPT-6 models. · Bugfixes - Preserved Bedrock DeepSeek model IDs, including ARN-based IDs, so they resolve
This week's release includes call hierarchy support, multi-select in the Git Panel, automatic language detection for untitled buffers, and project search on typ
Agent Zero Release Notes · Connector & Transport Hardening - Enforced negotiated WebSocket payload ceilings with a symmetric checksummed transfer protocol for l
Includes the fixes previously planned for 1.16.1, which was not published. The previous published version is 1.16.0. · Upgrade Notes - Cached responses use new
Changed: - Reuses shared conversation preparation, memory formatting, project/session/global scope utilities, and telemetry while preserving Pi's native extensi
Hotfix on top of 0.88.4. Combines 0.88.4-hotfix.1 and 0.88.4-hotfix.2. · - fix(worker): disable piece cache prewarm - Revert "fix(connections): stop oauth2 conn
Version 3.13.0 (September 8, 2026) · General · 1. eb18d77ca The .mjz encoder now writes the root file as model.xml in the archive as this is less susceptible to
What's Changed 🐛 Bug Fixes * fix(frontend): render no icon for plain select trigger by @Cristhianzl in https://github.com/langflow-ai/langflow/pull/14901 * fix:
What's Changed * fix: clean startup logs and initialize durable queue storage by @ashpreetbedi in https://github.com/agno-agi/agno/pull/10071 * fix: preserve Co
Open Interpreter 0.0.42 · Open Interpreter 0.0.42 improves provider and transport discoverability, keeps the command-line experience consistently branded, and a
✨ Features · - Support GPT-6 Astra models #11869 - Tool calling for goose-agent #11627 - Kotlin callers can now configure Databricks AI Gateway path #11631 · 🐛
New Models · - Qwen3.8-Flash-Next, a multimodal MoE model with 125B total parameters (~6B activated per token, including a 51B N-gram embedding table). It combi
The command-line tool for Vespa.ai. · Use it on Vespa instances running locally, remotely or in the cloud. Prefer web service API's to this in production. · See
This is the release note for v5.0.0. · Highlights · Major Enhancements to the Default Sampler · Optuna v5.0 introduces the first major update to its default sam
This release fixes a few bugs in version 6.14 and adds a couple of minor features that didn't make it into the previous release. For a full rundown of recently-
Use Ollama models in ChatGPT Desktop · Ollama models can now be used directly in ChatGPT Desktop, so you can keep your existing workflow while running open mode
Highlights · *786 PRs from 214 contributors.* · New models in this release (see the cookbook for all supported models): · | Model | Type | PRs | Cookbook | |---
Overview · llama.cpp 0.4.0 adds initial Qwen3.8-Flash-Next and Nemotron-3-Puzzle support, on-demand tensor reading, per-slot server context limits, video input
Change log · Improvements · * https://github.com/qdrant/qdrant/pull/10342, https://github.com/qdrant/qdrant/pull/10420 - Speed up and saturate memory bandwidth
🔥 What's New? · New Features · - #14215 - Scope integrations per expert and surface them in the UI (by @0ubbe) - #14216 - Surface agent work on the homepage via
Browser Use 0.13.10 · - Upgrade Browser Harness to 0.1.13. - Exact-pin all declared runtime, optional, development, and build dependencies. - Upgrade and migrat
Feature · * Add native pdf pipeline (#3979) (1c96dd3) * jats: Preserve external hyperlinks (#4029) (cf38d90) * iwork: Complete the parsing of Apple Pages docume
Minor Changes · - 9447932: Dereference internal $ref/$defs in tool input schemas before provider translation, so properties reachable only through a reference k
What's Changed · Security hardening (#280) This release hardens DiskANN against type confusion and malformed vector datums. Previously, the operator-class SQL r
MLflow 3.16.0 includes several major features and improvements · Major New Features · - 🎨 Custom Trace Views: Design your own trace UI in plain English — descri
Release Notes · v3.1.1 · Bug Fixes · - The OpenAI Chat Completions and Responses converters no longer raise on an assistant message with no content parts, which
Changes: · * d8ec44919265b3641b7898a02038cbfe14590ee5 .NET: Bump package version to 1.80.1 (#14370) * 3438d882147b6e3e19a161a5f0dcb64e23e181db .NET: update SDK
What's Changed * fix: vlm does not have chat template applied in accuracy mode with pr… by @vinovo in https://github.com/qualcomm/GenieX/pull/1428 · Full Change
Plane-4.7.1 was released as the stable version for planes and VTOLs on 04-Sep-2026 Release announcement: https://discuss.ardupilot.org/t/plane-4-7-1-released/14
Paperclip v2026.831.1 · > Released: 2026-09-02 · A patch on top of 2026.831.0 that fixes first-run onboarding. It carries a single targeted change; everything e
PyTorch 2.14.0 Release Notes · - Highlights - Backwards Incompatible Changes - Deprecations - New Features - Improvements - Bug fixes - Performance - Documentat
What's Changed 🚀 Features * Integrate DeepEPv2 by @irexyc in https://github.com/InternLM/lmdeploy/pull/4783 * feat(pytorch): support Kimi K2.6 by @qescccczmr in
What's New · New Tools · - browser_start_recording / browser_stop_recording (opt-in via --caps=devtools) — Record actions that the user performs manually in the
Added · - ♿ Accessibility mode reaches the menus. Accessibility mode now marks the menu entry you are pointing at and the model already chosen with a stronger b
🛠 Breaking Changes · - fix: make table existence manifest-authoritative by @BubbleCal in https://github.com/lancedb/lancedb/pull/3919 - refactor(python): requir
🎉 Crawl4AI v0.9.3 Released! · 📦 Installation · PyPI: · Docker: · Note: Docker images are being built and will be available shortly. Check the Docker Release wor
This patch release fixes a multi-vector loading bug: PyLate checkpoints that carry both a [Q]/[D] prefix and a text prompt lost the prefix, so they were encoded
Release: v2.26.2 [skip ci] Patch · - Improve: Native enums and PEP 604 unions (a1ca7c3) - Make: Ship a Python source distribution (d921aef) - Fix: Retry Linux f
koboldcpp-1.120 · - Minor fix to assistant gen prefills being triggered incorrectly - Fixed a bug where failsafe mode was incorrectly selected - Added support f
Added New Feature: 🔀 CivitAI Workflow Browsing - The Model Browser now browses and imports ComfyUI workflows from CivitAI — pick the new Workflows model type (o
Patch Changes · - #2842 f86e609 Thanks @seanmcguire12! - update the MCP SDK to v1.30 and refresh supporting dependency versions · - #2836 94a0ee4 Thanks @seanmc
What's Changed * chore(deps): bump claude-code-action to v1 by @william-u10d in https://github.com/Unstructured-IO/unstructured/pull/4448 * Added Sign up link f
Activity and Chain of Thought Overhaul · We overhauled our UI for long-horizon agentic tasks that show multiple thinking and tool calls to now roll up into a si