AI.info
AI Tools Updates — Page 76
What changed in the AI tools this week — dated, from the makers’ own changelogs.
- iFlytek Spark (Xinghuo) — 2.0.0
添加语音听写，语音合成，在线翻译，实时语音转写，语音转写五个在线能力；修复图片理解调用stop无法中断问题；SDK的清单文件中删除权限声明，权限声明以及获取放到demo或项目中去做；获取设备指纹逻辑中删除IMEI项
- neural.love — Introducing WorldModel v1: Revolutionary AI Photo Colorization
Analyzes the scene, recreates the world when the photo was taken, and handles scratches, tears, and other imperfections.
- Not Diamond Model Router — Not Diamond launches model routing on AWS marketplace
launches model routing on AWS marketplace
- Shortwave — Auto-reminders for emails with no reply ⏰
Enable automatic reminders to follow up on emails if there's no reply after a set time.
- neural.love — Introducing Video Generation: Transform Images into Motion with neural.love
Transform images into videos and create videos using text prompts.
- Stable Diffusion 3.5 — Replace Background & Relight
The SD3 and SD3.5 models now support the cfg_scale parameter.
- Stable Diffusion — Replace Background & Relight
Lets users swap backgrounds with AI-generated or uploaded images while adjusting lighting to match the subject.
- Shortwave — Shared read statuses 📨
Teammates on the same thread can now view email opens, even if they didn’t send the email.
- Stable Diffusion 3.5 — Stable Diffusion 3.5 Medium & Stable Image Ultra
The endpoint supports sd3.5-medium; Stable Image Ultra is now powered by Stable Diffusion 3.5 Large.
- Neutron 5 — System Requirements
VST2 format is no longer supported.
- InkubaLM — InkubaLM: A Small Language Model for Low-Resource African Languages
Lelapa AI introduces InkubaLM, a 400M parameter model designed to serve African communities without requiring extensive resources.
- neural.love — Introducing Stable Diffusion 3.5: The Next Evolution in AI Image Generation
Stable Diffusion 3.5 is integrated into neural.love and offers a base for development and fine-tuning.
- Stable Diffusion 3.5 — Stable Diffusion 3.5 Large & Large Turbo
The endpoint supports sd3.5-large and sd3.5-large-turbo in the model parameter.
- neural.love — Introducing Creative Upscale: Unleash AI-Powered Image Enhancement
Enhances images, adds creative elements based on image data, and lets users control AI-generated additions.
- Readwise — Export your highlights to NotebookLM 🤖🎙️
We worked directly with the NotebookLM team to build an integration for getting your highlights and notes into the system via Google Docs.
- Mubert Render — My Profile Gets a Major Upgrade!
Update payment methods, view transaction history, and upgrade or downgrade a subscription plan within your profile.
- Speak — Live Roleplays powered by OpenAI Realtime API
Live Roleplays powered by OpenAI Realtime API.
- neural.love — Introducing AI-Powered Background Removal and Replacement
The AI image editing suite includes Remove Background and Background Replacement.
- Thematic — Key Design Changes
Export Assistant groups data by score and country; Answers adds access links, pinned comments, and support for multiple datasets.
- Mubert Render — Linked Channels Management
Add, change, and remove links to connected channels like YouTube and Instagram.
- vidIQ — Find Winning YouTube Video Ideas Fast with vidIQ Remix
Take any video on YouTube and use it to create content for your channel!
- Chai-1 — Introducing Chai-1: Decoding the molecular interactions of life
A model for molecular structure prediction across tasks relevant to drug discovery.
- Retouch4me Heal — Update: Heal to version 1.200
New AI model detects moles and freckles with a slider to keep or remove them; updated interface; process photo sessions in Arams.
- HHEM (Hughes Hallucination Evaluation Model) — support GPU inference
- Thematic — Thematic Updates: Answers, Theme Highlighting & Snowflake!
Theme highlighting shows insights, scores and conversation references; Answers supports email sharing; Snowflake Reader pulls data from Snowflake.
- Ozone 12 — Compatibility Updates & Improvements
Fixed crash when loading certain presets or modifying crossover types; Added High DPI display support on Windows
- HHEM (Hughes Hallucination Evaluation Model) — remove unecessary `peft` dependency
- Dataiku LLM Mesh — New feature: Managed LLM fine-tuning
Managed LLM fine-tuning
- HHEM (Hughes Hallucination Evaluation Model) — update for pipeline
- Jais — G42 launches JAIS 70B and 20 other AI Models to Champion Arabic Natural Language Processing
JAIS 70B was released; 20 models, across 8 sizes, ranging from 590M to 70B parameters, are available.
- Not Diamond Model Router — Launching Not Diamond
- HHEM (Hughes Hallucination Evaluation Model) — updated files for hhem-2.1-open
- Ozone 12 — Compatibility Updates & Improvements
Added Logic 11 support
- Dataiku LLM Mesh — Major new feature: Multimodal embeddings
Multimodal embeddings
- Speak — Leveling up our core speech recognition systems at Speak
Leveling up speech recognition systems at Speak.
- Mubert Render — My Profile
Manage and delete your account, track monthly limits, change your password, and control newsletter subscriptions in one place.
- Remove.bg — Introduced new parameters: shadow_type and shadow_opacity . Deprecated parameter: add_shadow . Extended type parameter with new values: "graphic", "transportation" and "animal".
Introduced parameters shadow_type and shadow_opacity; deprecated add_shadow; extended type with values "graphic", "transportation" and "animal".
- Hyperscience Platform — Machine transcription of Korean multiline fields
We’ve fixed an issue that caused machine transcription to fail on multiline Korean fields.
- Trinka AI — Trinka Introduces LaTeX Proofreader for LaTeX Documents
This new tool proofreads LaTeX documents and aims at enhancing the writing and editing process for the academic community.
- Neutron 5 — Bug Fixes & Improvements
ADDED Support for macOS Sonoma, Cubase 13, Nuendo 13 and Reaper 7; CHANGED Minimum supported macOS version to macOS Monterey.