xAI's Grok 4.20 scores 48 on Artificial Analysis' Intelligence Index—6 points ahead of Grok 4 but trailing Gemini 3.1 Pro Preview and GPT-5.4 at 57. The model distinguishes itself with a 78% non-hallucination rate on the AA Omniscience test, the highest recorded across any model tested.
OpenAI released version 2.27.0 of its Python SDK on March 13, 2026, adding significant improvements to the Sora video generation API. The update introduces character API support, video extension and editing capabilities, and higher resolution export options.
AI2 has released MolmoSpaces and MolmoBot, robotics models trained exclusively in simulation that transfer directly to real robots without manual real-world data collection or fine-tuning. The approach eliminates months of teleoperated demonstrations typically required for simulation-trained robots. Both systems are open-source.
Hume AI has open-sourced TADA, a speech generation model that maps exactly one audio signal to each text token, achieving 5x faster processing than comparable systems. The model produced zero transcription hallucinations across 1,000+ test samples and runs on smartphones, available in 1B and 3B parameter versions under MIT license.
Perplexity AI has launched Personal Computer, a paid AI agent service priced at $200 per month that operates autonomously to handle emails, presentations, and application control. The service aims to provide continuous AI assistance for routine digital tasks without human intervention.
OpenAI is acquiring Promptfoo, an AI security platform that helps enterprises identify and remediate vulnerabilities in AI systems during development. Terms of the acquisition were not disclosed.
An SEC filing reveals Nvidia plans to spend $26 billion on open-weight AI models over the next five years. The investment targets the open-source gap left by OpenAI, Meta, and Anthropic while countering the rise of Chinese open-source models and deepening developer dependence on Nvidia hardware.
Perplexity launched Personal Computer, an AI agent tool that runs continuously on a local Mac with full access to your files and applications. The system operates as a "digital proxy" controlled remotely from any device, expanding on Perplexity's earlier Computer product announced last month.
Meta has unveiled four generations of custom-designed AI chips focused on inference workloads, aiming to reduce inference costs across its platforms serving billions of users. The move represents a significant step toward reducing Meta's dependence on GPU manufacturers like Nvidia and AMD.
NVIDIA has released Nemotron-3-Super-120B-A12B-NVFP4, a 120-billion parameter text generation model featuring a latent Mixture-of-Experts (MoE) architecture. The model supports 8 languages including English, French, Spanish, Italian, German, Japanese, and Chinese, and is available on Hugging Face with 8-bit quantization support through NVIDIA's ModelOpt toolkit.
NVIDIA has released Nemotron-3-Super-120B-A12B-BF16, a 120 billion parameter model designed for text generation and conversational tasks. The model employs a latent mixture-of-experts (MoE) architecture and supports multiple languages including English, French, Spanish, Italian, German, Japanese, and Chinese.
Google has released Gemini Embedding 2, its first native multimodal embedding model that represents text, images, video, audio, and documents in a unified vector space. The model eliminates the need for separate embedding models across different modalities in AI pipelines.
A study by research organization METR found that approximately 50% of AI-generated code solutions that pass the widely-used SWE-bench benchmark would be rejected by actual project maintainers. The finding exposes a significant gap between industry-standard code generation benchmarks and real-world code review standards.
OpenAI plans to integrate its Sora video generator as a built-in feature within ChatGPT, according to The Information. Currently available only on a standalone website and app, the integration would let users generate videos directly in the chatbot, similar to how image generation was added last year.
AI2 is developing physical AI systems trained primarily on virtual simulation data rather than expensive real-world demonstrations. The approach, demonstrated through projects like MolmoBot, addresses the historical bottleneck of manually collecting hardware training data.
OpenAI has released IH-Challenge, a training dataset designed to teach AI models to reliably distinguish between trusted and untrusted instructions. Early results show significant improvements in security and prompt injection defense capabilities.
An AI agent deployed by security firm Codewall gained full read and write access to McKinsey's internal AI platform Lilli within two hours without credentials or insider knowledge. The exploit used SQL injection, a decades-old vulnerability technique, to compromise a system serving over 43,000 employees for strategy work and client research.
Meta has developed four new custom AI chips called MTIA (Meta Training and Inference Accelerator) processors designed to power its AI models and recommendation systems. The move represents the company's ongoing effort to reduce dependence on Nvidia's expensive processors while managing massive compute requirements.
Hume AI has released TADA-1B, a 1 billion parameter text-to-speech model available on Hugging Face under an MIT license. The model, which combines speech and language capabilities, has already accumulated over 3,100 downloads since its January 12 release.
Qualcomm and Wayve announced a technical partnership to integrate Wayve's AI driving layer with Qualcomm's hardware platform for production-ready advanced driver assistance systems. The collaboration aims to accelerate autonomous vehicle innovation by combining hardware and software expertise.