AI Launch Radar

Two new open-source frameworks are positioning themselves as flexible alternatives to locked-in services like Claude Managed Agents.

• DeepSeek...

Agentic ESOpt uses evolution strategies instead of RL to fine-tune long-horizon LLM agents, requiring only inference-level GPU memory for...

Three new tools signal expanding infrastructure for coding agents.

• Frugal Tokens tracks session costs, cache misses, and model spend with detailed...

Amazon is automatically upgrading all compatible Fire TV devices in the U.S. to its AI-powered Alexa+ at no cost and without requiring Prime. The move...

Stripe's $7B+ acquisition of OpenRouter signals the rising strategic value of AI API routing as a core infrastructure layer. The leading LLM router's...

CSGHub is an open-source platform for managing LLM assets—models, datasets, spaces, and code—that deploys on-premise for secure, offline operation....

Two launches signal infrastructure for deploying agents beyond local setups.

• Cronloop runs scheduled AI loops (5 min to weekly) with durable memory...

Attackers are misusing AI coding agents for semi-autonomous intrusions, data theft, and ransomware.

• Chinese operators combined Claude Code with...

AgentR 3.0 now runs complete first interviews autonomously, structured and adaptive to counter AI-assisted cheating while basing shortlists on evidence from real work scenarios.

Cloud Platform Advancements

• 🔥 AWS AI Agents: Amazon Quick now lets anyone build and deploy autonomous AI agents in seconds using plain language...

TREX launches as an AI code reviewer that executes code to perform reviews, drawing 50 points of interest on Hacker News.

• GLM-5.2 adds 1M-token context and dual reasoning modes for coding tasks
• LoopCoder-v2 (7B) hits 64.4 on SWE-bench Verified, beating models 30x...

xAI just launched Grok Imagine 1.5, its new image-to-video model now widely available.

The release promises sharper realism, better physics, and...

• Enterprise advantage: Chat-first Copilot Projects pulls in Teams meetings, generates Excel breakdowns, and leverages existing OneDrive/Microsoft...

• Wipro partnership launches Claude Centre of Excellence in Bengaluru with plans to certify 10,000 employees over 18 months
• Seoul office opens to...

• Soofi releases transparent industrial AI model in Europe with full weights and training methodology docs
• Adam (YC W25) launches open-source AI...

Gemma 4 E2B running natively in-browser at 255 tok/s proves on-device AI can deliver practical performance without cloud dependency.

Major providers are expanding AI agents and governance features to meet enterprise demands.

• Databricks launched Genie Code agents for ML...

Regulatory Actions on Frontier Models

• 🔥 Anthropic Model Disablement: Anthropic disabled access to its Fable 5 and Mythos 5 models following a...

Z.ai's newly released GLM-5.2 delivers open-weights performance that beats GPT-5.5 on multiple long-horizon coding benchmarks at roughly 1/6th the API...