OpenAI

• Advanced Voice Mode: OpenAI has rolled out advanced voice mode to all Plus and Team users, except in Europe, the UK, Switzerland, Iceland, Norway, and Liechtenstein. This mode allows users to converse with ChatGPT using their smartphone for a more natural, human-like interaction. The advanced voice mode leverages GPT-4 for faster responses and can interpret text, vision, and audio inputs. For more details, visit the Advanced Voice Mode FAQ.

• Executives Departing: Shortly after the release of advanced voice mode, OpenAI CTO Mira Murati and two other executives announced their departure. Murati, after six and a half years with the company, decided to leave. Additionally, Barret Zoph, the company's Vice President of Research, announced his resignation, followed by Chief Research Officer Bob McGrew, who also praised the company's latest AI model in his farewell note.

• Sam Altman's Vision: OpenAI CEO Sam Altman shared more of his vision for an AI-powered future in a blog post titled The Intelligence Age.

Alibaba

• Dense, easy-to-use, decoder-only language models available in sizes ranging from 0.5B to 72B in both base and instruct variants.

• Pretrained on a large-scale dataset encompassing up to 18T tokens.

• Significant improvements in instruction following, long text generation (over 8K tokens), understanding structured data (e.g., tables), and generating structured outputs, especially JSON.

• Enhanced resilience to diverse system prompts, improving role-play implementation and condition-setting for chatbots.

• Context length support up to 128K tokens and can generate up to 8K tokens.

• Multilingual support for over 29 languages, including Chinese, English, French, Spanish, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, Arabic, and more.

Google

• Gemini Models Update: Google updated its production-ready Gemini models and introduced two new models: Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002. Updates include:
• Over 50% reduced pricing on 1.5 Pro (for prompts < 128K)
• 2x higher rate limits on 1.5 Flash and ~3x higher on 1.5 Pro
• 2x faster output and 3x lower latency
• Updated default filter settings

Anthropic

• Contextual Retrieval: Anthropic introduced Contextual Retrieval, significantly improving the retrieval step in Retrieval-Augmented Generation (RAG). It uses two sub-techniques: Contextual Embeddings and Contextual BM25, reducing failed retrievals by 49% and by 67% when combined with reranking. These improvements markedly enhance retrieval accuracy. Details can be found here: Introducing Contextual Retrieval.

• Revenue Growth: Anthropic's revenue is reportedly set to jump to $1 billion this year, representing 1,000% year-over-year growth. More details can be found here: Anthropic Revenue.

Meta

• Orion Augmented Reality Glasses: Meta introduced Orion, the first true augmented reality glasses, combining the look and feel of regular glasses with immersive AR capabilities. More details can be found here: Introducing Orion, Our First True Augmented Reality Glasses.

• Llama 3.2: Meta released Llama 3.2, which includes small and medium-sized vision LLMs (11B and 90B) and lightweight, text-only models (1B and 3B) suitable for edge and mobile devices, available in both pre-trained and instruction-tuned versions. More details can be found here: Llama 3.2: Revolutionizing Edge AI and Vision with Open, Customizable Models.

• New AI Features for Ray-Ban Meta Glasses: Meta added new AI features to Ray-Ban Meta glasses, helping users remember things like where they parked, translate speech in real-time, answer questions about what they see, and more. Details can be found here: Ray-Ban | Meta Glasses Are Getting New AI Features and More Partner Integrations.