Articles about "context window"

Chinese AI firm DeepSeek has released DeepSeek-V4, a large language model with context capabilities extending to one million tokens. The new model claims improved world knowledge and reasoning, and is available in two editions optimized for different hardware. Its launch comes as China faces increasing chip export restrictions from the United States.

Large language models such as ChatGPT have transformed the tech landscape since late 2022. Researchers are now developing advanced LLMs, or LLMs+, that aim to be more efficient, cost-effective, and capable of handling complex tasks for extended periods. New approaches in model architecture and improved 'working memory' are driving this progress.

The economics of multi-agent AI systems are now central to the viability of modern automated business workflows. NVIDIA's introduction of Nemotron 3 Super, a highly scalable agentic AI architecture, aims to tackle cost and efficiency challenges faced by enterprises deploying advanced automation. The architecture offers innovations in memory, compute efficiency, and workflow alignment for mission-critical industries.

Anthropic has launched Claude Sonnet 4.6, a new version of its midsize large language model, with a significant upgrade to a 1 million token context window. The model features enhanced coding, instruction-following, and achieves strong results on multiple AI benchmarks. Sonnet 4.6 now serves as the default model for both Free and Pro users.