In a week brimming with developments, leading AI companies have unveiled significant advancements in their models, navigated regulatory challenges, and forged strategic partnerships.
Google Introduces Gemini 2.5 Flash with “Thinking Budget”
Google has launched Gemini 2.5 Flash, an enhanced version of its AI model featuring a “thinking budget” tool. This innovation allows developers to control the computational reasoning of the AI, balancing performance with cost and response time. The model is now available in preview through the Gemini API via Google AI Studio and Vertex AI.
Anthropic Releases Upgraded Claude 3.5 Sonnet
Anthropic has unveiled Claude 3.5 Sonnet, boasting significant improvements in coding and tool use. The model has demonstrated substantial gains in performance benchmarks, particularly in the retail and airline domains. It is accessible via the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI.
OpenAI’s GPT-5 Delayed; GPT-4.1 Launched
OpenAI has announced a delay in the release of GPT-5, now expected later in 2025. In the interim, the company has launched GPT-4.1, focusing on enhanced coding capabilities. This move comes as OpenAI plans to phase out GPT-4.5 from its API by July 14, 2025.
Regulatory Scrutiny Intensifies for Google
The U.S. Department of Justice has initiated an antitrust trial against Google, alleging the company leverages its dominance in search and AI to maintain a monopoly. The trial has revealed that Google pays Samsung substantial sums to preinstall its Gemini AI app on devices, raising concerns about competitive practices.
Strategic Partnerships and Investments
- Reco Secures $25 Million: AI-powered cybersecurity startup Reco has raised $25 million in Series A funding, aiming to enhance security for cloud applications adopted without IT oversight.
- Mistral AI and CMA CGM Collaborate: French AI startup Mistral AI has partnered with shipping giant CMA CGM in a €100 million, five-year deal to improve customer service and media fact-checking using AI.
- AWS Offers Mistral’s Pixtral Large: Amazon Web Services has become the first major cloud provider to offer Mistral AI’s Pixtral Large model as a fully managed, serverless service within Amazon Bedrock.
