AI news for: Models And Releases
Explore AI news and updates focusing on models-and-releases for the last 7 days.
All (70)
13 news
50 posts
2 tools
5 videos
17
Oct
16
Oct
15
Oct
14
Oct
13
Oct
12
Oct
11
Oct
I tried the new Sora 2 to generate AI videos - and the results were pure sorcery
The new Sora 2 app can turn text or images into videos with dialog and sound effects in seconds....
TL;DR
OpenAI's Sora 2 allows users to generate short AI videos with impressive effects using text descriptions, and is currently available in the US and Canada by invitation only.
Key Takeaways:
Key Takeaways:
- OpenAI's Sora 2 can turn text prompts into short, AI-generated videos with a variety of styles and effects.
- The app and website allow users to edit, remix, or animate images directly within the Sora platform.
- Sora 2 is currently invite-only in the US and Canada, but will expand to other regions in the future, including the UK, EU, and Australia.
Skills for Claude will let you customize tasks with pre-set instructions - here's how
Available to Pro, Max, Team, and Enterprise users, the new capabilities shape Claude's output when carrying out routine or specialized tasks....
TL;DR
Anthropic has launched 'Skills for Claude', allowing users to create and upload custom instructions for the AI chatbot to follow.
Key Takeaways:
Key Takeaways:
- Users can create and upload their own custom skills for Claude to follow.
- Skills are essentially digital instruction manuals that make Claude more customizable and specialized.
- The launch of Skills marks a step towards making Claude more agentic, able to carry out complex tasks with minimal user oversight.
DeepSomatic, an open-source AI model, is speeding up genetic analysis for cancer research.
An overview of DeepSomatic, a new AI tool that helps identify complex genetic variants in cancer cells....
Anthropic turns to ‘skills’ to make Claude more useful at work
AI agents spent years as a concept and then as an experiment. Now, AI companies are devoting even more time and resources than before to make their ag...
TL;DR
Anthropic introduced Skills for Claude, a tool aimed at making its AI agent Claude more useful for end-users at work.
Key Takeaways:
Key Takeaways:
- Skills for Claude provides instructions, scripts, and resources to improve Claude's abilities for specific tasks.
- This feature is designed to reduce the time spent writing prompts and referring to past context.
- Box, Rakuten, Canva, and other companies have already used the tool, with Anthropic making it available to Pro, Max, Team, and Enterprise users.
Anthropic launches new version of scaled-down ‘Haiku’ model
Anthropic has released Claude Haiku 4.5, the newest version of its smallest model, billed as offering similar performance to Sonnet 4 "at one-third th...
TL;DR
Anthropic releases Claude Haiku 4.5, a lightweight AI model offering similar performance to Sonnet 4 at lower cost and higher speed.
Key Takeaways:
Key Takeaways:
- Haiku 4.5 reaches 73% accuracy on SWE-Bench and 41% on Terminal-Bench, matching Sonnet 4, GPT-5, and Gemini 2.5.
- The model's lightweight nature makes it suitable for deploying multiple agents in parallel and integrating with more complex models.
- Haiku 4.5 is expected to be particularly appealing for free versions of AI products and will support new styles of deployment in production environments.
You can test Microsoft's new in-house AI image generator model now - here's how
It's already scoring in the top ten at the AI leaderboard LMArena....
TL;DR
Microsoft has introduced its own in-house AI image generator model, MAI-Image-1, available for testing at the LMArena leaderboard site.
Key Takeaways:
Key Takeaways:
- Microsoft has been working on its own AI models in-house, shifting away from OpenAI's models and technology.
- MAI-Image-1 excels at generating photorealistic imagery and is currently ranked No. 9 on the LMArena leaderboard.
- The new model will be available in Copilot and Bing Image Creator soon and is currently available for testing at LMArena.
Related:
Microsoft AI announces first image generator created in-house
www.theverge.com
•
Oct 13, 2025
Is art dead? What Sora 2 means for your rights, creativity, and legal risk
OpenAI's Sora 2 gives anyone the power to make realistic AI videos - but what happens when creativity, copyright, and deepfakes collide in ways we can...
TL;DR
OpenAI's Sora 2 generative AI video creator raises legal and ownership risks, sparks debate on creativity and copyright law.
Key Takeaways:
Key Takeaways:
- The AI video tool raises real legal and ownership risks, with many creators and copyright holders concerned about the potential misuse of their work.
- The OpenAI CEO argues that Sora 2 supports creativity, but some critics disagree, highlighting the tool's ability to generate copyrighted content without proper permission.
- The use of AI video tools like Sora 2 also raises questions about the ownership and authorship of generated content, with some experts suggesting that human creators may be held liable for the AI's output.
This new Google Gemini model scrolls the internet just like you do - how it works
Now available in public preview, the new model is another step toward AI that can operate across web environments with minimal human oversight....
TL;DR
Google DeepMind has debuted a new AI model that can interact directly with website UIs, navigating a web browser just as a human would.
Key Takeaways:
Key Takeaways:
- The new Gemini 2.5 Computer Use model can execute tasks like clicking, typing, and scrolling directly within a web page.
- The model outperformed similar tools from Anthropic and OpenAI in terms of accuracy and latency, across multiple web and mobile control benchmarks.
- The new model comes with safety controls to prevent undesired actions, and is available now through the Gemini API in Google AI and Vertex AI.
Related:
Introducing the Gemini 2.5 Computer Use model
blog.google
•
Oct 08, 2025
Google’s latest AI model uses a web browser like you do
www.theverge.com
•
Oct 07, 2025
1Password says it can fix login security for AI browser agents
www.theverge.com
•
Oct 08, 2025
Google's new Gemini 2.5 Computer Use model can click, type, and scroll
www.zdnet.com
•
Oct 09, 2025
Claude now integrates directly with Microsoft 365
Here's what the new connector lets Anthropic's chatbot do, how it can benefit you, and who gets to access it....
TL;DR
Anthropic's Claude can now be accessed directly within Microsoft 365 to streamline collaboration and search capabilities.
Key Takeaways:
Key Takeaways:
- Claude can access SharePoint, OneDrive, Outlook, and Teams to pull information directly from those apps.
- The new 'enterprise search' feature allows businesses to integrate all critical apps for centralized resource retrieval.
- Admins must curate digital tools for the team-wide accounts, and the new features are available to Claude Team and Enterprise subscribers.
Google’s Photoshop-killer AI model is coming to search, Photos, and NotebookLM - Ars Technica
Google’s Photoshop-killer AI model is coming to search, Photos, and NotebookLM Ars Technica...
TL;DR
Google's Photoshop-killer AI model, Nano Banana, is expanding to search, Google Photos, and NotebookLM for conversational image editing.
Key Takeaways:
Key Takeaways:
- Nano Banana AI model is a 'major upgrade' over Google's previous image-editing model, providing more efficient conversational editing.
- The updated AI image editor will be available in search, Google Photos, and NotebookLM, bringing the feature to a wider user base.
- Nano Banana offers a range of video styles powered by AI, including whiteboard, anime, retro print, and more, available in NotebookLM.
Claude's latest model is cheaper and faster than Sonnet 4 - and free
Here's what Haiku 4.5 offers users and developers....
TL;DR
Anthropic launches its new small language model, Haiku 4.5, which is cheaper, faster, and safer than its predecessor while offering similar performance to the larger Claude Sonnet 4.
Key Takeaways:
Key Takeaways:
- Haiku 4.5 is faster and more cost-effective than Sonnet 4, costing one-third of the price and delivering twice the speed.
- Haiku 4.5 demonstrates competitive performance with Sonnet 4 and other large language models in various benchmarks, including coding, visual reasoning, and high school-level math.
- Haiku 4.5 has shown low rates of concerning behaviors and achieved an AI Safety Level 2 standard, making it Anthropic's safest model yet.
AI couldn't create an image of a woman like me - until now - BBC
AI couldn't create an image of a woman like me - until now BBC...
TL;DR
Recent updates to ChatGPT have led to improved image generation capabilities, including the accurate depiction of people with disabilities, but experts caution that AI bias remains a persistent issue.
Key Takeaways:
Key Takeaways:
- The latest updates to ChatGPT include improved image generation capabilities, such as the ability to accurately depict people with disabilities like arm prosthetics.
- Experts emphasize that AI bias is a pervasive issue and needs to be addressed through rigorous data testing and training methods that prioritize representation and diversity.
- The environmental impact of AI models, including energy consumption and data center usage, is also a growing concern and needs to be addressed.
Gemini 3.0 Pro is already referenced on Gemini's source code
If you still skeptical or think the screenshot is fake, here is a direct link to a gstatic JS source: [https://www.gstatic.com/\_/mss/boq-bard-web/\_/...
Trending AI Repos & Tools
Basically, it's a **CoreML/MLX translation of SimulStreaming** (2025 SOTA in simultaneous speech transcription), which itself is a combination Simul-W...
Qwen3-VL
14911Qwen3-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud....
Video Updates
17
Oct
16
Oct
15
Oct
14
Oct
13
Oct
12
Oct
11
Oct
Community talk
Meta just dropped MobileLLM-Pro, a new 1B foundational language model on Huggingface
GLM 4.6 is the new top open weight model on Design Arena
Anthropic just dropped Claude Imagine and it might be the biggest leap yet for Gen UI
New models Qwen3-VL-4b/8b: hands-on notes
GPT-OSS from Scratch on AMD GPUs
We built 3B and 8B models that rival GPT-5 at HTML extraction while costing 40-80x less - fully open source
new 1B LLM by meta
GLM 4.6 air when?
Major AI updates in the last 24h
Claude Haiku 4.5 Released
Introducing Claude Haiku 4.5: our latest small model.
Major AI updates in the last 24h
Qwen3-VL 4B vs 8B vs 235B
Qwen3-VL-4B and 8B Instruct & Thinking are here
Ring-1T open-source model released, achieving SOTA benchmark performance and silver-level IMO reasoning
AI highlights this week
GPT-5 Pro Tops FrontierMath Tier 4, Beating Gemini 2.5 Deep Think
A list of models released or updated this week on this sub, in case you missed any (10 Oct).
Claude Haiku 4.5 hits 73.3% on SWE-bench for $1/$5 per million tokens (3x cheaper than Sonnet 4, 2x faster)
PaddleOCR-VL, is better than private models
Google C2S-Scale 27B (based on Gemma) built with Yale generated a novel hypothesis about cancer cellular behavior - Model + resources are now on Hugging Face and GitHub
Gemini 3.0 Pro spotted
First Sora 2 video
My first 15 days with GLM-4.6 — honest thoughts after using Opus and Sonnet
[P] Nanonets-OCR2: An Open-Source Image-to-Markdown Model with LaTeX, Tables, flowcharts, handwritten docs, checkboxes & More
« In a few weeks, we plan to put out a new version of ChatGPT that allows people to have a personality that behaves more like what people liked about 4o »
Dolphin X1 8B (Llama3.1 8B decensor) live on HF
Llama5 is cancelled long live llama
Optimize my environment for GLM 4.5 Air
Creative writing statement from ChatGPT 5 introduction
GLM just blow up, or have I been in the dark?
ChatGPT and 4.o
Kwaipilot/KAT-Dev-72B-Exp model released
Now that is awesome
Just have a session this morning and Haiku 4.5 session limits feel significantly better, possibly 2x 2.5x Sonnet 4.5 in my estimates
Claude Haiku 4.5 Spotted
Claude’s file upload limit dropped from 6% to 4% — now I can’t work. Any workarounds?
Claude Sonnet 4.5 best AI for code generation, thoughts on worthy rivals?
Sharing a few image transcriptions from Qwen3-VL-8B-Instruct
GPT-OSS-20b TAKE THE WHEEL!
Label says 4o, voice feels 5 — started after yesterday’s app update. You seeing this too?
Kwaipilot/KAT-Dev-72B-Exp seems to be a great coding model？
Bless Claude 4.5
GLM 5 coming before the end of 2025
Looks like our automated overlords have arrived.
microsoft/UserLM-8b - Unlike typical LLMs that are 'assistant', they trained UserLM-8b to be the 'user' role
How is everyone's 4o memory been doing for the past few days?
Better 4o After the UI Update
They removed the model picker from the top! Why is no one talking about this??
Claude Code Context Window Issue