Blog

Zhipu AI's ChatGLM2 brings 32K context, 42% faster inference, and stronger coding capabilities in a compact 6B parameter package.

Technology Innovation Institute releases the groundbreaking 180B parameter Falcon 180B model, trained on 3.5T tokens and topping the Open LLM leaderboard.

Google's PaLM 2 represents a quantum leap in language modeling with 340 billion parameters, delivering breakthrough performance in multilingual capabilities, reasoning, and coding.

MosaicML's MPT-7B delivers enterprise-grade performance with Apache 2.0 licensing, making it the ideal choice for commercial applications requiring long context processing.

BigCode's StarCoder emerges as a powerful open-source alternative to proprietary code LLMs, trained on 1TB tokens across 80+ programming languages.

Stability AI releases StableLM, an open-source 7B parameter language model trained on 1.5 trillion tokens with permissive licensing for commercial use.

Discover how Vicuna 13B revolutionizes open-source AI by achieving 90% of ChatGPT's performance through innovative fine-tuning techniques.

Anthropic launches Claude 1, their first public language model featuring Constitutional AI safety measures and a massive 100K context window.

Discover how GPT-4 transformed AI with its multimodal capabilities, 1.8T parameter architecture, and unprecedented reasoning abilities that set new industry standards.