With just 3B active parameters, it matches Gemini-2.5-Pro and GPT-5-High, and even surpasses them on ChartQA and DocVQAval What can it achieve with only 3B active parameters?
For example, its “Thinking with Images” feature lets it zoom in, analyze details, and reason through
INNOVATION
-
3B Parameter Model Matches GPT-5 and Gemini Performance
By
–
-

Qwen Image Edit Plus LoRA Explorer Now Available
By
–
Qwen Image Edit Plus (2509) LoRA Explorer is here!
— Replicate (@replicate) 12 novembre 2025
Use any HF adapter fast.
Lightning fast.
Thanks to the @PrunaAI + @Alibaba_Qwen teams for making all of this possible pic.twitter.com/Vq99FWalzsQwen Image Edit Plus (2509) LoRA Explorer is here!
Use any HF adapter fast.
Lightning fast. Thanks to the @PrunaAI + @Alibaba_Qwen teams for making all of this possible -
GPT-5.1 Reddit AMA and Customization Updates Announced
By
–
Reddit AMA on GPT-5.1 and our customization updates. Tomorrow, 2PM PT.
-

Lowe’s Launches Comprehensive AI Strategy with Nvidia Dell
By
–
1,750+ stores. Millions of customers. One AI strategy. @Lowes partnered with @nvidia and @Dell for its comprehensive AI strategy with the goal to revolutionize customer shopping experiences, empower store associates, and transform merchandising. Read the full story:
-
Improving Frontend Design Through Skills Development
By
–
Blog post: https://
claude.com/blog/improving
-frontend-design-through-skills
… S/o @rgb_prithvi for the work here! -

Eliminating Purple LLM-Slop Design with Prompting and Claude Skills
By
–
Turns out you can eliminate the purple LLM-slop design aesthetic with good prompting + claude skills. We wrote a short post on it with more details:
-

SAS Microsoft Trusted AI Innovation at MSIgnite
By
–
Learn how SAS + Microsoft help you innovate confidently with trusted AI at #MSIgnite. Register for the theater session on Nov. 20: http://
2.sas.com/60167n9v6 -

GPT-5.1 ‘Thinking’ and ‘Instant’ versions launching
By
–


BREAKING : OpenAI is rolling out GPT-5.1 to ChatGPT users as per multiple reports. GPT-5.1 comes with "Instant" and "Thinking" versions. "GPT‑5.1 Thinking varies its thinking time more dynamically than GPT‑5 Thinking."
-
AI Privatization Threatens Research and Global Participation
By
–
What are we losing when AI knowledge becomes privatized? – Cross-pollination of ideas
– Reproducibility of research
– Global participation
– Talent pipeline This is where academia can play a major role, say @StanfordHAI leaders. #TeamOpenScience -
Tiffany Farriss discusses UN Open Source Week and digital sovereignty
By
–
Tiffany Farriss appeared on the most recent episode of the Talking Drupal podcast to discuss the UN's Open Source Week, digital public infrastructure, and digital sovereignty.