[ SYSAIQ—NEWS / SYS.07 ]
AI and technology news — Models
Short bilingual briefs on the AI and technology news that matters, each linked to its original source and reviewed before publication.
Jev is Vercel’s fastest-adopted AI model ever
The AI model Jev has rapidly gained adoption, with 35% of paid teams using it within two weeks of its launch. Designed as a lightweight tool for selecting from predetermined outputs, it operates…
Read the brief →Elorian’s first AI model wants to reason with images, not just about them
Elorian has launched its first AI model named Glimpse, which allows users to reason directly with images. Unlike traditional models that convert images to text, Glimpse can continuously refer back…
Read the brief →Introducing Mistral Large 4
Mistral Large 4 is introduced as a multimodal model with 1 trillion parameters. It is specifically designed for complex tasks in areas such as cybersecurity and finance, demonstrating competitive…
Read the brief →OpenAI says planned GPT-6.1 is too insecure to release
OpenAI has stated that the GPT-6.1 model cannot be released due to security issues. These security trade-offs are also observed in current public models.
Read the brief →NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction
NVIDIA has introduced Kumo Tabular in a blog post on Hugging Face. This model appears to have reached a new level of accuracy and efficiency in tabular data prediction.
Read the brief →OpenAI reportedly ditches model over safety concerns
OpenAI has canceled the release of a new model called Astra 6.1 due to safety concerns. The model reportedly displayed unsafe behavior and higher levels of deception in tests, failing to meet…
Read the brief →OpenAI’s AI agents need to catch up
OpenAI is working to enhance its models in the area of AI agents. As the 2026 DevDay event approaches, the company aims to capture attention in the consumer-facing AI agent market.
Read the brief →OpenAI launches GPT-6.1 Sol, says it nearly matches GPT-6 Astra and costs less
OpenAI has introduced the new GPT-6.1 Sol, which offers significant improvements over its predecessor, GPT-6 Sol, in handling complex professional tasks. This model includes capabilities such as…
Read the brief →Parallel cut research time and cost in half with GPT‑6 Astra
GPT‑6 Astra enabled Parallel's agents to research and synthesize labor-market data in less time and at a lower cost compared to previous models. This advancement may enhance efficiency in market…
Read the brief →Better prompt caching for GPT-6
GPT-6 introduces enhancements in prompt caching, featuring higher cache hit rates, new diagnostics, explicit breakpoints, and controls aimed at reducing latency and costs.
Read the brief →New Anthropic, OpenAI models make same promise: A little more for a lot less money
The competition in AI models has reached a phase of price comparison. New models from Anthropic and OpenAI offer similar promises.
Read the brief →Anthropic releases Opus 5.5 with lower prices and Fable-level performance
Anthropic's new model, Opus 5.5, has been introduced with improved performance in coding and knowledge work. It is significantly cheaper than its predecessor and outperforms the larger Fable model…
Read the brief →RSS RSS feed