Google Launches Gemini 4 Argon for Browser-Based 3D Generation
Google has released Gemini 4 Argon, an AI model capable of generating interactive 3D browser environments from user prompts, with API pricing starting at $2 per million input tokens.

Google has released Gemini 4 Argon, an AI model capable of generating interactive 3D browser environments from user prompts, with API pricing starting at $2 per million input tokens.

Developers on the OpenRouter platform are increasingly choosing cost-effective non-frontier AI models, which now account for 84% of all processed tokens.

Qwen AI platform has released the Qwen-Image-2.1 Pro and Turbo APIs, enabling developers to integrate image generation and editing tools directly into their applications.

OrcaRouter has released the OrcaCyber Zero 1.5 cybersecurity model, featuring a 1 million token context window and specialized training for security research and vulnerability analysis.

Microsoft launched Microsoft-Decision-1, a specialized Qwen3.5-9B based model for decision scoring, offering high-speed probability outputs for software integration at $0.042 per million tokens.

OpenAI has released the Decisions API in public beta, offering typed evaluations for text and images with a 10x speed improvement over existing endpoints.

Xiaomi is phasing out its MiMo-V2.5 AI models in favor of the newer V2.6 series, with a full transition deadline set for October 21, 2026.

A federal judge is weighing whether President Trump can charge up to $100 000 monthly for API access to his Truth Social posts amid allegations of constitutional violations.

Anthropic has released the Claude Haiku 5.5 model, featuring a 75% reduction in operating costs and new customizable performance settings for high-volume tasks.

A developer explores integrating the Luna Decisions API into an n8n real estate monitoring pipeline to improve price change classification and decision-making.

TypeSafe has introduced Jev, a specialized AI model for classification and scoring that avoids text generation to eliminate hallucinations and reduce latency.

Google released Nano Banana 2.1, an AI model for image generation and editing that offers improved performance and a 50% reduction in API pricing compared to the previous version.

San Francisco startup Monid has raised $7.7 million to build a unified marketplace for AI agents to discover and pay for third-party API tools.

Anthropic is offering startups a free year of its Claude Team plan and $1 000 in API credits as part of an expanded support program.

Developers have created an automated voice booking system for beauty salons using Python, Flask, and Google Calendar API to manage appointments without staff involvement.

A developer shares insights on building an MCP server, highlighting that while coding takes an evening, publication, authentication, and testing require significant long-term effort.

Reddit is phasing out RSS feeds and public API access by March 2027 to prevent unauthorized data scraping, while launching a $1 million migration fund for developers.

OpenAI has unveiled the Decisions API, a new tool for AI agents that delivers fast, constrained choices using a specialized GPT-6 Luna model.

A comparative analysis of 11 AI models shows that tokenization differences make processing Russian text up to 3.9 times more expensive on some platforms.

Reddit will terminate RSS support and restrict API access to curb unauthorized AI data scraping and boost revenue from data licensing.
