⏰ Urgent: Xiaomi MiMo-V2-Flash Free Window Closes Tomorrow

The most time-sensitive item in the collector today: Xiaomi’s MiMo-V2-Flash free window on OpenRouter closes tomorrow (2026-06-17). At standard pricing the model is already among the cheapest long-context options on the platform ($0.10/M prompt · $0.30/M completion), but the free slot makes the next 24 hours the obvious sandbox target.

🤝 Anthropic Heads to the White House Over Mythos Suspension

The Anthropic-vs-US-government story enters a new phase: Anthropic has agreed to meet with the Trump administration to discuss the suspension of Claude Fable 5 and Claude Mythos 5, and the cybersecurity community is now openly pushing back on the underlying ban.

  • BBC: Anthropic to meet White House over AI tool suspension — the meeting frames the dispute as a negotiation rather than a fait accompli.
  • CNBC: Anthropic to meet with Trump administration over Mythos dispute — confirms the agenda is the Mythos-tier access restriction.
  • Reuters / BBC: Cybersecurity leaders are publicly urging the US to lift the curbs on Anthropic’s security models.
  • TechCrunch: Cybersecurity vets protest ‘dangerous’ US government ban on Anthropic’s most powerful models.
  • The Guardian: Anthropic to disable its most advanced AI models after US order limiting foreign access.
  • The New York Times: U.S. Bars Foreigners From Using Anthropic’s Most Advanced A.I. Models.
  • WSJ (scoop): Amazon CEO’s talks with US officials triggered the crackdown on Anthropic’s models.
  • Reuters: Amazon voiced concerns about Anthropic AI models before US crackdown, source says.

The WSJ/Reuters cluster points to AWS’s own security review as the proximate trigger, which raises the awkward question of whether the ban is rooted in an inter-cloud rivalry as much as in national-security concerns. Watch the White House meeting for any softening of the foreign-access restriction.

💰 Salesforce Acquires Fin for $3.6B — Largest Agentic CX Deal To Date

Salesforce announced it will buy AI customer-service platform Fin for $3.6 billion, the largest pure-play agentic CX deal to date. Fin (formerly part of the Intercom ecosystem) lets enterprises deploy AI agents that handle end-to-end customer support workflows.

  • CNBC: Salesforce to buy AI customer service platform Fin for $3.6 billion to boost agentic offerings.
  • TechCrunch: Salesforce acquires AI customer service platform Fin for $3.6 billion.
  • Reuters: Salesforce deepens AI automation push with $3.6 billion Fin buyout.
  • CMSWire: Salesforce Agrees to Buy Fin for $3.6 Billion, Largest Agentic CX Deal to Date.
  • qz.com: Salesforce acquires Fin, formerly Intercom, for $3.6 billion.

The acquisition is the clearest signal yet that “agentic” workflows — AI agents acting on behalf of customers inside a CRM — have graduated from feature to market category. Expect Microsoft, Google, and ServiceNow to respond with their own CX-agent acquisitions in the next two quarters.

🚀 Cerebras IPO Pops 108% — First Huge Tech Offering of 2026

Cerebras Systems raised $5.5 billion in its IPO and the stock popped 108% on debut, making it the first huge tech offering of 2026 and validating the AI-chip public-market thesis after a slow start to the year.

  • TechCrunch: Cerebras raises $5.5B, then stock pops 108%, in the first huge tech IPO of 2026.
  • IndexBox: Cerebras Systems IPO Raises $5.5 Billion in One of Largest Semiconductor Offerings.

The pop is meaningful beyond Cerebras itself: it reopens the IPO window for AI-infrastructure companies (CoreWeave, Groq, others reportedly waiting) and gives the public market a fresh valuation anchor for non-Nvidia AI silicon.

📊 The AI Layoff Wave Becomes a Powder Keg

A cluster of major employers publicly blamed AI for workforce reductions this week, and the pattern is now drawing systemic attention. The common thread: companies are cutting headcount even as revenue hits record highs, framing AI as the productivity multiplier that obviates the cut roles.

  • TechCrunch (analysis): The AI layoff wave is becoming a powder keg — frames the trend as politically and economically unsustainable.
  • Cloudflare — AI made 1,100 jobs obsolete; revenue hit a record high. (TechCrunch, Fortune, Yahoo Finance)
  • Cisco — cuts nearly 4,000 jobs in an AI-focused restructuring as orders surge. (TechCrunch, Reuters, CFO Dive)
  • Block / Jack Dorsey — halving employee base on the thesis that “most companies will do the same.” (TechCrunch, CNN, Fortune, Forbes, CNBC — Block shares jumped as much as 24% on the announcement.)
  • X — simultaneously kicked off a rebuilt, AI-powered ad platform. (TechCrunch, Social Media Today)

The Dorsey move is the most aggressive and the most-watched: a public CEO laying off 40% of staff and explicitly saying every other company will do the same. The market reaction (24% pop on Block) is the strongest possible endorsement of the “AI-first headcount” doctrine — and the political risk for every other CEO who delays.

🇮🇳 Sarvam Becomes India’s Newest AI Unicorn at $1.15B

Indian AI startup Sarvam raised $234 million led by HCLTech, with Bessemer Venture Partners and others joining. The round pushes Sarvam past the $1B threshold to become India’s newest AI unicorn.

  • TechCrunch: Sarvam becomes India’s newest AI unicorn with $234 million funding round led by HCLTech.
  • Business Standard: Sarvam AI becomes unicorn after raising $234 mn; HCLTech acquires over 10%.
  • Moneycontrol: Sarvam AI turns unicorn after raising $234 million from HCLTech, Bessemer Venture Partners, others.
  • Swarajya: Sarvam AI becomes India’s newest unicorn after raising £180 million at £1.15 billion valuation.
  • Analytics India Magazine: Sarvam Becomes Unicorn After $234 Mn Funding Round Led by HCLTech.

The HCLTech-as-lead structure is notable: an Indian IT-services giant (rather than a US VC) anchoring a sovereign-AI startup is a template India is likely to repeat as it builds out an Indic-language AI stack.

🔐 NewCore Raises $66M for AI Agent Identity

A new entrant, NewCore, emerged from stealth with $66M in seed funding to build “security-first identities for AI agents” — the digital passports needed as AI agents start acting as employees inside enterprise systems.

  • TechCrunch: As AI agents become employees, NewCore emerges with $66M to give them identities.
  • SiliconANGLE: NewCore launches security-first identities for AI agents after closing $66M seed funding round.
  • GovInfoSecurity: NewCore Launches With $66M to Rebuild Identity for AI Agents.
  • PR Newswire: NewCore Emerges from Stealth with $66M to Rebuild Workforce Identity for the Agentic Era.

The launch timing — same week as Block’s “AI agents as employees” announcement and the Salesforce-Fin deal — is not coincidental. Identity, access control, and audit trails for AI agents are now a category, not a feature.

🛰️ In Brief

  • Adobe launches Firefly AI Assistant for Photoshop — its first creative agent, with announced integration with Anthropic’s Claude. (TechCrunch, Reuters, Forbes, Adobe) Early reviews from Digital Camera World: “It feels like an overly enthusiastic intern.”
  • Meta launches ‘AI Mode’ on Facebook — a new chat-style surface that pulls from public content across Meta’s platforms. (TechCrunch, Firstpost, Meta Newsroom)
  • SpaceX is now public — post-IPO update — the stock gained 19% in its first full day of trading; the record $75B IPO will turn ~4,400 employees into millionaires per NYT estimates. (TechCrunch, CNBC, NPR, NYT, Reuters, Fortune)
  • OpenAI quietly confirms S-1 filing — confidential IPO paperwork is now with the SEC, one week after Anthropic’s similar filing. (MLQ.ai)
  • OpenAI launches the OpenAI Partner Network — a formal channel for systems integrators, cloud partners, and consultancies to deliver OpenAI-based solutions.
  • A satellite autonomously learned to find things on its own — early data from an on-orbit foundation-model experiment. (Bitcoin World, TechCrunch)
  • Musk defends AI ambitions as SpaceX IPO reveals trouble — commentary on the tension between SpaceX’s capital needs and the regulatory risk around xAI/Grok. (Free Malaysia Today)

🎁 Other Limited-Time Free Models

Three additional free-preview windows remain open on OpenRouter alongside Xiaomi’s closing slot:

  • Z.ai: GLM 4.5 — 131K context. Free until 2026-06-19 (2 days left). Standard $0.60/M prompt · $2.20/M completion.
  • Meta: Llama 3 70B Instruct — 8K context. Free until 2026-06-19 (2 days left). Standard $0.51/M prompt · $0.74/M completion.
  • Anthropic: Claude Opus 4.6 (Fast) — 1M context. Free until 2026-06-29 (12 days left). Standard $30/M prompt · $150/M completion.

🤖 New Models on Hugging Face

Ten new checkpoints landed on the Hub in the past 24 hours, spanning Kimi K2.5 distillation (CharyZeng), Qwen 3.5/3.6 quantization experiments (ReliquaryForge, groxaxo), RL/plasticity training runs (iamseungpil, Cisco1963, timaeus), a robotics endovascular policy (txmedai), and small-scale GPT-2 mid-training experiments.

🔥 Trending GitHub Repositories

Five repos made today’s trending board — the standout is a satirical AI-coding agent that has racked up more stars than most serious frameworks.

  • DietrichGebert/ponytail — “Makes your AI agent think like the laziest senior dev in the room.” JavaScript ⭐ 17,289
  • omnigent-ai/omnigent — Meta-harness unifying Claude Code, Codex, Pi, and other agents. Python ⭐ 2,001
  • SkyBlue997/enableMacosAI — One-click unlock of full Apple Intelligence on China-region Macs (macOS 27 / Apple Silicon). Shell ⭐ 1,295
  • orange2ai/renwei-writing — “人味儿写作” — an AI agent skill for editing writing without erasing the human voice. ⭐ 684
  • coder/boo — GNU screen-style terminal multiplexer built on libghostty. Zig ⭐ 606

📊 Today’s totals: 10 HF models · 1 OpenRouter model · 5 trending repos · 4 limited-time free models · 20 items · 14+ external news stories.

Reading time: ~8 minutes. Coverage window: 2026-06-15 evening to 2026-06-16 evening UTC.