Google is adding Veo 3.1 Lite to Ultra subscriptions at zero credit cost starting May 10, 2026. The model costs less than half of Veo 3.1 Fast but generates videos at the same speed according to Google, though quality tradeoffs remain unclear.
April 13, 2026
AWS has introduced Lambda-based reward functions for Amazon Nova model customization through reinforcement fine-tuning (RFT). The serverless architecture automatically scales from 10 concurrent evaluations per second during experimentation to 400+ during production training, supporting both objective RLVR and subjective RLAIF approaches.
Microsoft is testing OpenClaw-style autonomous agents for 365 Copilot that would run continuously to complete tasks on behalf of users, according to The Information. The company plans to demonstrate some features at its Build conference on June 2nd.
OpenRouter has released Elephant Alpha, a 100B-parameter text model with a 256K context window and 32K output token limit. The model is available at no cost through OpenRouter's platform, supporting function calling, structured output, and prompt caching.
Google released Gemma 4, an open-weights model family that runs directly on edge devices with multi-step planning and autonomous workflow capabilities. The Apache 2.0 licensed model bypasses traditional cloud security controls by executing entirely on local hardware, creating a governance blind spot for enterprise security teams.
Anthropic released a Claude add-in for Microsoft Word, completing its integration across all three major Office applications. The Word add-in joins existing Excel and PowerPoint add-ins, allowing Claude to rewrite text, respond to comments, and track changes across Office documents.
Anthropic has launched Mythos, an AI model the company claims can identify and exploit zero-day vulnerabilities with significant capability. The model has not been released publicly, with Anthropic citing security concerns. The announcement raises questions about the model's actual capabilities versus pre-IPO positioning.
U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned bank executives this week and encouraged them to test Anthropic's newly announced Mythos model for detecting security vulnerabilities. According to Bloomberg, major banks including Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are already testing the model alongside JPMorgan Chase, despite Anthropic's stated plan to limit initial access.
OpenAI's pricing page created confusion about usage limits across its new ChatGPT Pro tiers. An OpenAI employee clarified that the $100 plan provides at least 10x Plus usage while the $200 plan offers at least 20x, though both currently benefit from a temporary 2x boost expiring May 31.
Open-weight models from Google, Alibaba, Microsoft, and Nvidia have crossed a threshold from research projects to enterprise-grade systems. The shift reflects a growing divide: frontier models from OpenAI and Anthropic are too expensive and pose data security risks for most enterprises, while open alternatives now deliver sufficient capability at a fraction of the cost.
A large-scale study testing 34,198 real-world skills reveals that AI agent performance drops drastically when moving from curated benchmarks to realistic conditions. Claude Opus 4.6 saw pass rates fall from 55.4% with hand-selected skills to 38.4% in truly realistic scenarios, while weaker models like Kimi K2.5 actually perform below their no-skill baseline.
Arcee AI has released Trinity-Large-Thinking, a 400-billion-parameter open-weight reasoning model with a mixture-of-experts architecture that activates only 13 billion parameters per token. The model matches Claude Opus 4.6 on agent benchmarks like Tau2 and PinchBench but lags on general reasoning tasks. The company spent approximately $20 million—roughly half its total venture capital—to train the model on 2,048 Nvidia B300 GPUs over 33 days.
MiniMax has released MiniMax-M2.7, a 229-billion parameter model that uniquely participates in its own evolution during development. The model achieves 66.6% medal rate on MLE Bench Lite and 56.22% on SWE-Pro benchmarks, with native support for multi-agent collaboration and complex tool orchestration.