Releases

MLflow AI Gateway provided a unified interface for routing requests to multiple commercial and open model providers.

Databricks introduced automated and human evaluation tooling for assessing generative-AI quality on enterprise data.

Databricks announced managed vector search integrated with the lakehouse for retrieval-augmented generation and semantic search.

Databricks introduced an AI assistant for generating, explaining and repairing SQL and Python with organizational context.

Databricks launched Lakehouse AI, later branded Mosaic AI, as an integrated platform for building generative-AI applications on enterprise data.

Midjourney released V5.2 with improved aesthetics, shorter prompts, and Zoom Out.

MosaicML released the MPT-30B base, instruct and chat models with commercial-use terms.

ElevenLabs expanded its generative voice tools with Voice Design.

ElevenLabs launched a classifier for detecting audio generated by its models.

ElevenLabs raised a $19 million Series A led by Nat Friedman, Daniel Gross, and Andreessen Horowitz.

Microsoft Research introduces Phi-1, showing that curated textbook-quality data can produce strong code and reasoning performance in a compact model.

Microsoft Research introduces Kosmos-2, linking generated language to object regions in images through grounding.

Meta introduced Voicebox, a generative speech model capable of editing, sampling and cross-lingual speech generation.

CoreWeave released Tensorizer, an open-source tensor serialization tool that streams models from S3, HTTP, and file storage to reduce PyTorch model startup time.

Oracle announced a managed generative-AI service developed with Cohere for enterprise use cases.

Oracle selected Cohere models and OCI infrastructure to deliver native generative-AI services across Oracle applications.

Mistral AI raised a €105 million seed round, one of Europe's largest seed financings, at a reported valuation of roughly €240 million.

Cohere raised $270 million from Inovia, Oracle, NVIDIA, Salesforce and other investors to scale enterprise generative AI.

Microsoft Research introduces Orca, a 13-billion-parameter model trained from rich explanation traces produced by larger models.

Palantir began using intensive AIP Bootcamps to move customers rapidly from use case to working AI deployment.

Hugging Face launched a daily research-paper discovery feed connected to community profiles, models, datasets, and discussions.

Perplexity introduced its paid Pro subscription with expanded searches and access to advanced models.

NVIDIA announced DGX GH200, connecting 256 Grace Hopper Superchips into a large shared-memory AI system.

Hugging Face launched an open and reproducible benchmark leaderboard for comparing community language models.

Perplexity introduced Copilot, a guided multi-step research mode later renamed Pro Search.

Perplexity launched its native Android application.

Baidu began transforming Wenku into an ERNIE-powered document and creativity platform.

Microsoft announces a common plugin model spanning ChatGPT, Bing, Dynamics 365 Copilot and Microsoft 365 Copilot.

Anthropic raised $450 million led by Spark Capital to scale reliable AI products and safety research.

AWS launched Inferentia2-powered EC2 instances for large-scale generative-AI inference.

Meta released MMS speech-recognition and synthesis models covering more than 1,100 languages.

Meta unveiled its first in-house AI inference accelerator, designed initially for recommendation workloads.

Anthropic and Zoom announced a partnership to use Claude in Zoom products, alongside an investment from Zoom Ventures.

Anthropic expanded Claude's API context window from 9K to 100K tokens.

Scale launched EGP as a model-agnostic enterprise platform for deploying secure generative-AI applications with proprietary data, fine-tuning, evaluation, and red teaming.

Scale launched Donovan, a secure decision-support platform using large language models, retrieval, and mission data for government and defense users.

Scale deployed Donovan for the U.S. Army XVIII Airborne Corps, describing it as the first large-language-model platform deployed on a classified network for the Corps.

Google Cloud launches Model Garden to discover, customize and deploy Google and third-party foundation models.

Google launches PaLM 2, improving multilingual, reasoning and coding capabilities across Google products and Cloud.

The release of Replit Code v1 3B established Replit's native family of open coding language models.

Replit released its open 3-billion-parameter code model for code completion.

IBM Consulting launched a center with more than 1,000 specialists to support enterprise generative-AI adoption.

IBM Consulting announced a dedicated generative-AI center and delivery practice alongside watsonx.

IBM announced watsonx as an integrated AI and data platform for enterprise foundation-model development and governance.

Meta released ImageBind, a joint embedding model spanning images, text, audio, depth, thermal and IMU data.

Anthropic published the principles used to train and guide Claude's behavior through Constitutional AI.

MosaicML released MPT-7B, Base, Instruct, Chat and StoryWriter variants under commercially usable licenses.

BigCode released StarCoder, a 15.5B-parameter code model trained on permissively licensed code from The Stack.

Midjourney released V5.1 with stronger default aesthetics and coherence.

Databricks agreed to acquire data-governance company Okera to strengthen governance and AI security capabilities.