Three Weeks, Four Models, One Verdict

Four launches landed between July 8 and 9: OpenAI opened its GPT-5.6 family to general availability, xAI took Grok 4.5 public as a cheap Cursor-trained coding model at $2 / $6 per 1M tokens, Meta shipped Muse Spark 1.1 as its first paid model at $1.25 / $4.25, and ByteDance released Seedream 5.0 Pro , according to FelloAI. The timing was not coincidence. It was collision.

GPT-5.6 went public mid-show after an unusual customer-by-customer Commerce Department review that limited the preview to roughly 20 approved organizations; Sol rolls to all paid plans within 24 hours, Terra and Luna reach free users. Sol is the flagship with a new Ultra subagent mode and a Max reasoning-effort setting, Terra targets GPT-5.5-level quality at half the cost, and Luna is the fast tier , ThursdAI reported. Elon Musk's SpaceXAI launched Grok 4.5, its smartest model yet, and its first release since going public and acquiring the AI coding startup Cursor , Axios confirmed. Meta shipped Muse Spark 1.1 on July 9, 2026, its most capable model yet for real-world coding and agentic tasks, and started charging developers to use its own model for the first time through the new Meta Model API. Pricing is $1.25 per 1M input and $4.25 per 1M output tokens, with $20 in free credits for every new account .

The week's real drama, though, played out before any of those models shipped.

Can the Government Pull a Model Offline?

On Friday, June 12, the US government applied export controls to Anthropic's newest models, Claude Fable 5 and Claude Mythos 5. This required the company to restrict access to foreign nationals, whether inside or outside the United States , Anthropic disclosed. Because the order took effect immediately and the company had no reliable way to verify nationality in real-time, it suspended access to both models for all users .

The trigger was a jailbreak. The export control directive on June 12 came after the government became aware of a report in which Amazon researchers had found a method of bypassing Fable 5's safeguards: prompting it so that it identified a number of software vulnerabilities. In one case, the model produced code demonstrating how the relevant vulnerability could be exploited . Anthropic's testing confirmed that many less capable models—including Claude Opus 4.8, GPT-5.5, and Kimi K2.7—could identify the same vulnerabilities as Fable 5 did in the report. When it came to the demonstration of how to exploit the single vulnerability, every model tested could produce the same demonstration as Fable 5 .

As of June 30, the export controls on Fable 5 and Mythos 5 were lifted , and Fable 5 became available starting July 1 to users globally on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork , according to the company. The three-week blackout cost Anthropic an estimated $965 billion valuation hit and enterprise deployments across 15 countries, per industry analysis.

The precedent is the story. A June 2 executive order created a voluntary path for companies to have frontier models reviewed before release. Fable 5 never went through that path. The government reached for export controls instead. That is the tell: when Washington wants to move fast on a frontier model, it still has no binding process, only improvised ones , The Hacker News observed.

Why Is Google Losing Its Best People?

On Thursday, Noam Shazeer announced he was leaving to go to OpenAI. Shazeer is the famed AI researcher who helped build Google's earliest LLM-based chatbot system, LaMDA, in 2021, and then left in frustration when the internet giant dragged its feet in commercializing it , Fortune reported. Google licensed Character's technology for a reported $2.7 billion payment to bring him back in 2024. He left again less than two years later.

Just days after the Shazeer news, Google DeepMind researcher John Jumper announced he too was leaving—in this case, to join Anthropic. Jumper shared the 2024 Nobel Prize in chemistry with Google DeepMind CEO Demis Hassabis for their work creating AlphaFold, the AI system that could predict the shape of proteins from their DNA sequences, solving a 50-year grand challenge in biochemistry .

Jumper doesn't strike as the kind of person who is primarily motivated by money, and he wouldn't leave Google DeepMind unless he thought the scientific opportunity at Anthropic was actually better—which is much worse news for Google DeepMind. Industry watchers are wondering aloud whether the AI lab is slipping back from the lead pack in the AI race. Its top AI models, Gemini 3.5 Flash and Gemini 3.1 Pro, are often ranked outside the top five places on various AI benchmark leaderboards , Fortune noted.

Google DeepMind has delayed the release of its Gemini 3.5 Pro AI model to July 17, 2026, scrapping the existing 2.5 Pro architecture for a complete rebuild. The overhaul targets improvements in mathematical reasoning, SVG scene generation, and image quality to compete with OpenAI's GPT-5.6 and Anthropic's Fable 5 , according to BigGo Finance.

The departures matter because researchers of this caliber rarely move. The most coveted researchers offer more than technical knowledge contained in papers or code. They bring judgment about which ideas to pursue, experience running enormous experiments and the ability to recruit other sought-after scientists. For top researchers, deciding where to work often involves a complicated calculus: potential financial rewards, access to computing power, each company's prospects of leading the field, and whether its leadership will wield that power responsibly , Axios explained.

Where Is the Money Actually Going?

AI research labs now represent roughly $1.68 trillion of aggregate private valuation. And 93.6% of that $1.68 trillion sits in a single bucket: frontier generalists, with OpenAI, Anthropic, xAI, and Safe Superintelligence absorbing almost all of it , NEA reported in a July 8 analysis. OpenAI was reported by market coverage to have finalized a $110 billion funding round at a $730 billion valuation , according to startup funding trackers.

AI Startup Funding news, July, 2026 shows one hard truth: AI money is still huge, but most of it is going to U.S. startups, giant rounds, and a small circle of companies. Crunchbase data says nearly 88% of AI startup funding in 2026 went to U.S.-based companies, while mega-rounds for OpenAI, xAI, and Anthropic shaped market attention .

The concentration is deliberate. Every dollar that flows into a frontier generalist at hundreds of billions in valuation assumes that their next model will deliver another step change in model performance to temporarily capture a market held by a competitor in an endless tug of war. VCs are not underwriting today's models. They are assuming that scale, plus a breakthrough no one has seen, will manifest into market creation and capture , NEA's analysis concluded.

This week, most of the largest U.S. startup funding rounds centered around artificial intelligence. This was true for the week's largest venture financing, a $1.5 billion Series F for AI inference technology provider Baseten, as well as a majority of rounds in the Top 10 , Crunchbase reported. Shield AI secured $1.5 billion in Series G funding, part of a broader $2.25 billion capital package, valuing the company at $12.7 billion, up 140% in just one year. The startup builds the Hivemind autonomous pilot platform, which was selected by the U.S. Air Force for its Collaborative Combat Aircraft program. The company projects $540M+ in 2026 revenue, representing 80% year-over-year growth .

What Changed This Week

The voluntary pre-release review framework became real. Anthropic's three-week blackout proved the government can and will pull a frontier model offline, even when weaker models exhibit the same capability. OpenAI's staggered GPT-5.6 rollout—20 approved organizations first, then general availability—shows labs are now negotiating deployment timelines with Washington, not just announcing them. Google lost two researchers who define what frontier AI research looks like, and the market repriced Alphabet down 5-6% in response. Meta pivoted from open-weight Llama to closed-weight Muse Spark, ending a four-generation run of releasing model weights to the public.

What to Watch

Google's Gemini 3.5 Pro ships July 17 with a rebuilt architecture—watch whether the delay translates to competitive performance or just catches the lab up to where OpenAI and Anthropic already are. Anthropic's new HackerOne program for Fable 5 jailbreak reporting will test whether crowdsourced red-teaming can keep pace with adversarial prompt engineering. The Commerce Department's 30-day pre-release review window, established by the June 2 executive order, comes due July 2—the process document will clarify whether frontier labs face binding review or voluntary consultation. And xAI's claim that it will release "completely trained from scratch" foundation models roughly every month for the rest of 2026 will either prove compute-capacity leadership or become another missed Musk timeline.