AI.info
AI Tools Updates — Page 94
What changed in the AI tools this week — dated, from the makers’ own changelogs.
Em inglês
- Capacities — Recent continuous improvements
- vLLM — v0.31.0rc4
Fixed MTP acceptance collapse under FULL graphs.
- Retool Agents — Create PostgreSQL and REST API resources with natural language
The app building agent can create REST API or PostgreSQL resources from a prompt.
- Retool Agents — New OpenAI models available in Retool
Retool now supports GPT 6 Luna, GPT 6 Sol, GPT Image 2.5 Flare, and GPT Image 2.5 Sunburst for AI resource queries.
- Profound — Meet the new Projects, integrate more context, and more in Profound
Profound introduced Projects, workspaces for marketing initiatives with context, conversations, outputs, and AI Marketer suggestions.
- OpenHands Cloud — openhands: v0.76.0
Updated the OpenHands image tag to 1.68.1.
- OpenHands Cloud — openhands: v0.75.0
Updated agent-server and OpenHands image tags, and provisioned the KV store secret for Replicated installs.
- Sigma — Assistant in build mode: reuse data model metrics
Sigma Assistant in build mode preserves data model sources you attach and reuses their metrics instead of recreating tables and calculations.
- Mintlify — Desktop app improvements, agent comment replies, and more
Added desktop app improvements and agent comment replies.
- CodeRabbit — GitLab webhook migration | Self-managed GitLab
A migration script moves existing CodeRabbit project webhooks to a new endpoint while preserving hook IDs and event settings.
- Neon — More storage for your 100 Free plan projects release - Oct 02, 2026
Neon Free plan project storage increased from 0.5 GB to 1 GB.
- Unsloth — Command Palette + Desktop UI/UX
Added a command palette and shareable run settings, sped up Laya decisions, expanded hosted Decision API support and improved LoRA training.
- Baseten — Baseten CLI 1.1.0
You can now rename a model from the terminal with baseten model rename. The model keeps its ID, endpoints, and deployments.
- Novelcrafter — Model Updates and AI Improvements
Refreshed system model collections, added a search modal, and enabled choosing one-off models for Chat threads.
- LangFlow — v1.13.0.dev30
Langflow nightly v1.13.0.dev30
- Make — Google Gemini AI video module update, TypeSafe, and more app updates
Explore Google Gemini AI video model changes, a new TypeSafe app, WhatsApp Business Cloud pricing, and Claude Sonnet 5.5 support.
- Retool — Create PostgreSQL and REST API resources with natural language
The app building agent can create REST API or PostgreSQL resources from prompts in the app builder.
- Retool — New OpenAI models available in Retool
Retool supports GPT 6 Luna, GPT 6 Sol, GPT Image 2.5 Flare, and GPT Image 2.5 Sunburst for AI resource queries.
- Tonic Structural — v1884
Bug fixes and other internal updates.
- Midjourney — Alpha Changelog - 10/1/26
Launched style previews in the sidebar, larger images in Create, and folder-level default parameters on alpha.
- Hugging Face — Open-sourcing AstaBrief, the fast report-generation model in Asta
AstaBrief, a fast report-generation model in Asta, is open-sourced.
- Runway — Seedance 2.5 Draft Mode in Agent
Agent can now generate with Seedance 2.5 Draft mode, creating 480p drafts with post-generation enhancement.
- Runway — Ideogram 4.5
Ideogram 4.5 is now supported in Runway MCP, bringing its image generation capabilities directly into your MCP workflows.
- Ollama — v0.35.1
Ollama supports Clef and Clef Flash through /v1/systemone, with requests that include images alongside text state.
- DeepEval — python-v4.2.8
- Braintrust — October 2026
Added SQL-query charts with visual editing and Windows DPAPI credential storage shared across terminal and MSIX apps.
- ON1 Photo RAW — September 2026 ON1 Recap: Cleaner High-ISO Shots, Photo RAW 2027 & More
ON1 NoNoise AI 2027 launched as a standalone app and plugin.
- Raycast — 🎛️ Performance Improvements and Auto Model
Raycast uses less memory and CPU, and Auto Model now offers Lightweight, Balanced, and Flagship preferences.
- Tembo — Session Forking, Claude Subscriptions, and New Models
Fork a session with its conversation context and a workspace snapshot; connect Claude subscriptions; GPT-6.1 Sol and Claude Sonnet 5.5 are available.
- Letta — Release v0.34.2
Added opt-in automatic reply relay and tripled the generated subagent name pool.
- Harvey AI — Accessibility Improvements: Screen Readers and Keyboard Navigation
Screen reader and keyboard navigation improvements are now live across the Harvey web app.
- Harvey AI — 130+ New Legal Research Sources
We've increased the number of data sources available to provide the latest relevant jurisdictional information.
- Factory Droids — CLI v0.232.0, Desktop v0.189.0: Legacy model warnings, session sidebar sorting, faster MCP connections, plus Windows and Slack fixes
Model pickers show discount savings and warn about legacy models; sessions sort by status, and HTTP MCP connections start faster.
- Weights & Biases — v0.85.1
Audit logs now include details about deleted runs.
- Enterprise h2oGPTe — v1.7.5
Adds agent skills and an LLM registry, rewrites guardrail categories, gives Prompt Guard a new backend, and fixes scheduling and Graph RAG issues.
- Meticulous — Meticulous crawl covers more ground
It now accepts multiple start URLs in a single run, and adds a new depth mode that prioritises interacting with page elements.
- Meticulous — Agents can grow your coverage
Sessions an agent records to cover new code can now be added straight to your selected set, with coverage updating immediately.
- Meticulous — Agents can approve diffs
Agents can approve or ignore expected diffs themselves, clearing the Meticulous check fully autonomously without a human review.
- Meticulous — Accessibility checks are now fully out of beta
Violations are traced right to the React component that rendered them, plus WCAG AA/AAA rule presets and check reports.
- Fundraise Up — Suggested amounts and impact descriptions in Tap to Donate and Giving Station
Tap to Donate and Giving Station campaigns can suggest donation amounts and show an impact description for each amount.