Model Performance

Twitter

Facebook

Copy Link

Key Facts

Key PlatformsLMSYS Arena, Open LLM Leaderboard

Key BenchmarksMMLU, HumanEval, GSM8K, ARC

Top News

Carbon Robotics unveils Large Plant Model trained on 150M+ plants to transform farming

Seattle-based Carbon Robotics launched its Large Plant Model, an AI model that detects and identifies plants instantly without retraining. Trained on over 150 million photos from 100+ farms across 15 countries, the model powers LaserWeeder robots that use lasers to eliminate weeds, helping farmers reduce labor costs and herbicide use while improving yields.

Technology2 sources

Carbon Robotics unveils Large Plant Model trained on 150M+ plants to transform farming

Technology2 sources

AI creativity matches average humans but top creators still outperform machines, new studies show

Two major studies reveal a nuanced picture of AI creativity versus human creativity. While generative AI scored better than average participants in a 100,000-person creativity experiment, the top 10% of humans far exceeded AI performance. Separately, CMU research found AI-generated music was judged less creative than human compositions, highlighting the complex relationship between AI and humans in creative fields.

Science2 sources

AI creativity matches average humans but top creators still outperform machines, new studies show

Science2 sources

AI mammography screening cuts aggressive breast cancer diagnoses by 12% in landmark trial

A groundbreaking Swedish trial involving over 100,000 women shows AI-supported mammography screening reduces interval cancers by 12% and detects 29% more cases than standard screening. The MASAI trial demonstrates AI can support radiologists in early detection while cutting workload by 44%, without increasing false positives.

Health7 sources

AI mammography screening cuts aggressive breast cancer diagnoses by 12% in landmark trial

Health7 sources

OpenAI Retires GPT-4o Model on Feb 13, Pushing Users to GPT-5.2 Despite Loyal Following

OpenAI announced it will retire GPT-4o and several other ChatGPT models on February 13, just two weeks after the announcement. Despite the model's beloved conversational style and warmth, OpenAI says only 0.1% of users still choose it daily, representing about 800,000 people. The company previously reversed this decision after intense user backlash but now feels confident that GPT-5.2 and new personalization features can satisfy longtime fans.

Technology13 sources

OpenAI Retires GPT-4o Model on Feb 13, Pushing Users to GPT-5.2 Despite Loyal Following

Technology13 sources

Google DeepMind's AlphaGenome AI tackles the 98% of human genome that doesn't code for proteins

Google DeepMind unveiled AlphaGenome, a deep-learning AI model that predicts how mutations in non-coding DNA affect gene expression across sequences up to one million base pairs long. The tool outperforms existing models in 25 out of 26 tasks and is already being used by nearly 3,000 scientists worldwide to investigate rare diseases, cancer mutations, and design new gene therapies.

Science14 sources

Google DeepMind's AlphaGenome AI tackles the 98% of human genome that doesn't code for proteins

Science14 sources

Gemini 3 Flash's Agentic Vision uses code execution to inspect images step-by-step

Google DeepMind introduced Agentic Vision in Gemini 3 Flash, transforming static image analysis into an active investigative process. The model uses Python code execution to zoom, annotate, and manipulate images through a Think, Act, Observe loop. This approach delivers a 5-10% quality boost across vision benchmarks and is now available through the Gemini API in Google AI Studio and Vertex AI.

Technology4 sources

Gemini 3 Flash's Agentic Vision uses code execution to inspect images step-by-step

Technology4 sources

Ai2 releases SERA open-source coding agents that adapt to any codebase for just $400

The Allen Institute for AI launched SERA, the first in its Open Coding Agents family, challenging proprietary AI coding tools. Built with just 32 GPUs by a small team, SERA solves over 55% of tough real-world coding problems on SWE-Bench while costing only $400 to reproduce—100 times cheaper than existing approaches.

Technology2 sources

Ai2 releases SERA open-source coding agents that adapt to any codebase for just $400

Technology2 sources

Adobe Photoshop upgrades Generative AI tools with 2K resolution and geometry-aware features

Adobe Photoshop has unveiled significant updates to its Firefly-powered generative AI editing tools, introducing 2K resolution output and geometry-aware capabilities. The enhancements to Generative Fill, Generative Expand, and Remove tools promise sharper detail with fewer artifacts. Creators also gain access to new non-destructive adjustment layers and a beta version of Dynamic Text for curved typography.

Technology3 sources

Adobe Photoshop upgrades Generative AI tools with 2K resolution and geometry-aware features

Technology3 sources

Google AI Overviews Upgrade to Gemini 3 with New Follow-Up Question Feature

Google is transforming how users interact with Search by upgrading AI Overviews to Gemini 3 and introducing the ability to ask follow-up questions directly within search results. The new feature seamlessly transitions users from quick AI-generated summaries into deeper AI Mode conversations, marking a shift from traditional link-based search to a more conversational, chatbot-like experience.

Technology19 sources

Google AI Overviews Upgrade to Gemini 3 with New Follow-Up Question Feature

Technology19 sources

Sam Altman announces dramatic hiring slowdown at OpenAI as AI efficiency reshapes workforce needs

OpenAI CEO Sam Altman revealed plans to dramatically slow hiring as AI tools enable smaller teams to accomplish more work. Speaking at a town hall event, Altman emphasized avoiding aggressive hiring followed by layoffs, while acknowledging ChatGPT's writing performance declined due to focus on coding and math capabilities in GPT-5.2.

Business4 sources

Sam Altman announces dramatic hiring slowdown at OpenAI as AI efficiency reshapes workforce needs

Business4 sources

Scientists Launch DinoTracker, an AI App That Identifies Dinosaur Footprints in Seconds

Researchers from Helmholtz-Zentrum Berlin and the University of Edinburgh have developed DinoTracker, a free AI-powered app that can identify which dinosaur made a footprint by analyzing its shape. The system matches human expert classifications about 90% of the time and has already uncovered intriguing clues about bird evolution that could push back their origins by tens of millions of years.

Science5 sources

Scientists Launch DinoTracker, an AI App That Identifies Dinosaur Footprints in Seconds

Science5 sources

OpenAI launches dedicated team to position AI as scientific collaborator across research fields

OpenAI has established a new AI for Science team led by Kevin Weil to help scientists accelerate research across mathematics, physics, chemistry, and biology. The company reports that 1.3 million weekly users now send 8.4 million messages on advanced science topics, representing 47% growth over the past year. GPT-5.2 achieves 92% accuracy on graduate-level benchmarks, though questions remain about long-term validation.

Technology4 sources

OpenAI launches dedicated team to position AI as scientific collaborator across research fields

Technology4 sources

AI Agents Hit Mathematical Wall: Vishal Sikka Paper Shows LLMs Can't Handle Complex Tasks

A paper by former SAP CTO Vishal Sikka and his son mathematically proves that LLMs powering AI agents cannot reliably execute tasks beyond a certain complexity threshold. The research challenges industry promises about autonomous AI systems, showing that transformer-based language models have fundamental limitations that even reasoning models can't overcome.

Technology5 sources

AI Agents Hit Mathematical Wall: Vishal Sikka Paper Shows LLMs Can't Handle Complex Tasks

Technology5 sources

AI agents score below 25% on workplace readiness test, exposing critical gaps in office work

Training-data company Mercor released the Apex-Agents benchmark, testing leading AI models on real professional tasks from law, investment banking, and consulting. Even top performers like Gemini 3 Flash and GPT-5.2 achieved only 24% and 23% accuracy respectively. The research reveals AI agents struggle most with multi-domain reasoning and synthesizing scattered information across workplace tools.

Science3 sources

AI agents score below 25% on workplace readiness test, exposing critical gaps in office work

Science3 sources

Inferact raises $150M to commercialize vLLM and build next-gen AI inference infrastructure

The creators of vLLM have launched Inferact with $150 million in seed funding at an $800 million valuation, led by Andreessen Horowitz and Lightspeed Venture Partners. As AI shifts from training to deployment, technologies that make AI inference faster and more affordable are attracting significant investor attention. The startup plans to enhance the open-source project while building commercial infrastructure.

Startups3 sources

Inferact raises $150M to commercialize vLLM and build next-gen AI inference infrastructure

Startups3 sources

Prev 1…101112…19 Next