⏰ Limited-Time Free Models
Several models are nearing their free access expiration on OpenRouter:
| Model | Expires | Pricing | Context |
|---|---|---|---|
| Baidu: Qianfan-OCR-Fast | May 28 (4 days) | $0.68/M tokens prompt, $2.81/M tokens completion | 65K |
| Mistral: Mistral 7B Instruct v0.1 | May 30 (6 days) | $0.11/M tokens | 4K |
| Google: Gemini 2.0 Flash Lite | Jun 1 (8 days) | $0.075/M tokens prompt, $0.30/M tokens completion | 1M |
| Google: Gemini 2.0 Flash | Jun 1 (8 days) | $0.10/M tokens prompt, $0.40/M tokens completion | 1M |
🚀 Alibaba Launches Qwen3.7-Max: 35-Hour Autonomous AI Agent
Alibaba’s Qwen team has released Qwen3.7-Max on OpenRouter, a reasoning agent model with a 1-million-token context window. The model can run autonomously for up to 35 hours and supports over 1,000 tool calls in a single session, including external harnesses like Anthropic’s Claude Code, according to a VentureBeat report. Priced at $2.5/M tokens (prompt) and $10/M tokens (completion), it ranks as one of the highest-performing Chinese AI models on the Chatbot Arena leaderboard.
🥽 Google’s AI Glasses: “Almost There” — TechCrunch
Google’s augmented reality smart glasses, co-developed with Samsung on the Android XR platform, received an early hands-on review from TechCrunch calling them “almost there.” The glasses were a highlight of Google I/O 2026, alongside Gemini and Search announcements. Key features include AI-powered real-time information overlay, navigation, and notification management. CNET recommends waiting for the Google model, while PCMag offered an early look at what works and what’s missing.
🏛️ Silicon Valley Helped Block Trump’s Expected AI Executive Order
The Washington Post and Politico report that pressure from major tech companies successfully stalled a planned Trump administration executive order on AI. Trump had expressed “many concerns” about the draft policy order, which would have mandated federal oversight of frontier AI model development. The development marks a significant reversal — the administration previously resisted AI regulation. A broader Fortune analysis notes the Trump administration is now “suddenly embracing AI oversight ideas it once rejected.”
💰 OpenAI, SpaceX Prep for Mega-IPOs
OpenAI is preparing a confidential IPO filing that could value the company at over $1 trillion, according to Axios and Fortune. SpaceX and Anthropic are also racing to go public in what CNBC analysts call a potential “market top” signal. The Wall Street Journal reports OpenAI is aiming for a speedy IPO process. The trio of AI-driven IPOs could reshape capital markets, with some analysts estimating combined valuations exceeding $2 trillion.
💼 Microsoft AI: More Expensive Than Human Workers?
A Fortune investigation reveals that Microsoft’s internal reports show AI-powered tools are sometimes more expensive than paying human employees to do the same work. This comes alongside Microsoft AI chief’s prediction that white-collar office jobs could be automated within 18 months. The reports highlight a growing tension between AI adoption economics and workforce displacement concerns as enterprises grapple with real-world AI deployment costs.
🤖 xAI Introduces Grok Build 0.1: AI Coding Agent
xAI has launched Grok-Build-0.1 on OpenRouter, its first AI coding agent in early testing for subscribers. The model features a 256K context window and is priced at $1/M tokens. Elon Musk demoed plugins for Microsoft Office integration, signaling xAI’s entry into the competitive AI coding assistant space currently dominated by GitHub Copilot, Claude Code, and Cursor.
🤗 Trending on Hugging Face
New models on Hugging Face this week include a diverse range of fine-tuned transformers, LoRA adapters, and research checkpoints. Notable uploads include juergengunz/fluxer (6 likes), pantomiman/qwen3.5-0.8b-grpo-musique-blackwell (a GRPO fine-tune with research paper reference), shaikhsalman/ZabaanAI-Urdu-3B (SFT-tuned Urdu language model), and saliacoel/chars (2 likes).
⭐ GitHub Trending: AI Edition
| Repository | Stars | Description |
|---|---|---|
| Doorman11991/smallcode | 1,209 | AI coding agent optimized for small LLMs — 87% benchmark with 4B-active model |
| datawhalechina/Agent-Learning-Hub | 1,080 | AI Agent learning resources and knowledge hub |
| LiuMengxuan04/shushu-internship-tool | 436 | Job description → project → resume → interview pipeline tool |
| lynote-ai/humanize-text | 435 | Free open-source AI text humanizer for undetectable AI-to-human writing |
| AbhishekK130804/Claude-Mythos-AI-Anthropic-App | 425 | Claude AI app interface (note: marketing-heavy description) |
💡 Key Trends
- AI Agents Go Autonomous: Alibaba’s 35-hour autonomous agent and xAI’s coding agent launch signal an industry-wide shift from chat-based AI to persistent, tool-using agents.
- Regulatory Pendulum Swings: Trump’s stalled AI order and sudden embrace of oversight highlights the volatile policy landscape.
- The AI Cost Reality Check: Microsoft’s internal cost comparisons and the Goldman Sachs CEO’s op-ed suggest the ROI of enterprise AI deployment is under increasing scrutiny.
- IPO Supercycle Approaching: OpenAI, SpaceX, and Anthropic going public in the same window could be the defining capital markets event of the decade.