Google has updated its Gemini 2.5 Pro and Flash models with several new features, including a Deep Think mode for 2.5 Pro, and native audio output, computer use, and thinking budgets for both models.
Notion has launched "Notion AI for Work," an integrated suite that embeds AI meeting notes, enterprise search, and document auto-drafting capabilities directly into its workspace platform, offering comprehensive AI functionality at a fraction of the cost of separate tools.
Granola has launched Granola 2.0, transforming its AI meeting assistant into a collaborative team workspace. The startup also announced it has secured $43 million in Series B funding in a round led by NFDG.
Salesforce is acquiring Convergence.ai, a pioneering AI agent company, to enhance its Agentforce platform with technology that enables AI to navigate complex digital workflows with human-like adaptability.
Startup Fastino has raised $17.5 million led by Khosla Ventures to develop Task-Specific Language Models that are faster, cheaper, and more efficient than general-purpose LLMs. Fastino aims to challenge several industry standards by offering its TSMs through a fixed-rate subscription system.
FutureHouse, the Eric Schmidt-backed nonprofit with ambitions to build an "AI scientist" within the next decade, has released a new tool called Finch aimed at accelerating biological discovery. Currently in the closed beta phase, Finch can assist with open-ended and directed data analysis tasks.
WisdomAI has secured $23 million in funding to launch the industry's first "Agentic Data Insights Platform," which uses AI agents powered by a "Knowledge Fabric" to transform business intelligence into an intelligent system that actively delivers insights without requiring specialized skills.
Mistral AI has launched Mistral Medium 3, an affordable, smaller model that delivers 90% of Claude Sonnet 3.7's performance at 8x lower cost. Medium 3 also outperforms competitors like Llama 4 Maverick in human evaluations, and can be deployed on any cloud environment with at least four GPUs.
The Qwen team has released Qwen3, a new family of open-weight language models comprising two MoE and six dense models. The Qwen3 model family features hybrid thinking capabilities, support for 119 languages, and competitive performance.
Microsoft has introduced three new small language models—Phi-4-reasoning, Phi-4-reasoning-plus, and Phi-4-mini-reasoning—that reportedly deliver complex reasoning capabilities comparable to much larger models while maintaining efficiency for deployment across various computing environments.
AI2's new Olmo 2 1B model outperforms similar-sized offerings from Google, Meta, and Alibaba across multiple key benchmarks, including GSM8K, TruthfulQA, and DROP, while maintaining a small enough size to run on resource-limited hardware.
Amazon Nova Premier, now generally available on Amazon Bedrock, can handle complex workflows requiring multiple tools and data sources. It also features outstanding model distillation capabilities, enabling users to transfer its advanced features to smaller, faster models for production deployment.