AI News July 9, 2026: GPT-5.6 Launches Publicly, Anthropic Extends Fable 5, Meta Ships Muse Image
Today in AI: OpenAI officially launches GPT-5.6 Sol, Terra, and Luna to the public after getting US government clearance — the biggest model release of the summer. Anthropic fires back by extending free Fable 5 access through July 12. Meta ships Muse Image with Arena-topping quality and invisible watermarking. CNBC confirmed that 30–46% of enterprise AI token usage now flows to Chinese models. And Samsung posts a record quarter that eclipses its entire 40-year chip history.
GPT-5.6 Sol, Terra, and Luna Launch Publicly — OpenAI’s Biggest Release of 2026
OpenAI confirmed on July 8 that GPT-5.6 Sol, Terra, and Luna will be publicly available starting Thursday, July 9. The company posted on X: “We’re expanding preview access globally now.” Sam Altman followed with: “GPT-5.6 sol launches thursday! happy building.” The three-model family was first previewed on June 26 for a small group of trusted partners, after the Trump administration requested a pre-release government review.
According to Engadget and Axios, the US Department of Commerce gave OpenAI clearance for a wider rollout after additional testing. Pricing is set at $5/$30 per million input/output tokens for Sol, $2.50/$15 for Terra, and $1/$6 for Luna. Sol Ultra scored 91.9% on Terminal-Bench 2.1 — the highest of any model. Terra matches GPT-5.5 performance at half the cost, making it the likely everyday workhorse for most developers. Expect access to roll out by product and account tier rather than all at once.
Anthropic Extends Free Fable 5 Access to July 12 — Timed to Counter GPT-5.6
Hours before Claude Fable 5 was set to leave paid subscriptions on July 7, Anthropic reversed course. The company announced on X: “We’re extending access to Claude Fable 5 on all paid plans through July 12.” Pro, Max, Team, and premium Enterprise subscribers can continue using the model for up to 50% of their weekly usage limit at no extra cost. After July 12, Fable 5 moves to usage-credit billing at $10/$50 per million input/output tokens.
The timing is impossible to miss. Neowin reported that Anthropic’s extension came “just as OpenAI appears to be preparing a broader release of GPT-5.6.” With both frontier models competing for developer mindshare in the same week, the extension is a calculated retention play. Anthropic has said it hopes to restore Fable 5 as a standard subscription benefit “once capacity allows” but hasn’t shared a timeline.
Meta Launches Muse Image — #2 on Arena, With Invisible Watermarking
Meta launched Muse Image on July 7, a new AI image generation model from its Superintelligence Labs. The model ranked #2 on Arena for text-to-image generation, single-image editing, and multi-image editing based on human-preference Elo rankings as of July 5. Muse Image powers the Meta AI app and will roll out to Instagram, WhatsApp, Messenger, and Ray-Ban Meta AI glasses in the coming weeks.
Capabilities include photo restoration, room redesign using Facebook Marketplace products, text rendering inside images, and the ability to @mention Instagram accounts in prompts to use their public photos. Every image carries Content Seal, Meta’s invisible watermarking system designed to survive cropping, compression, and screenshots. Meta also previewed Muse Video, which ranked #3 on Arena for text-to-video. With $125–145 billion in 2026 capex committed to AI infrastructure, Meta is finally shipping consumer-facing products at a pace that matches its spending.
Chinese AI Models Now Account for 30–46% of US Enterprise Token Usage
CNBC published a major investigation on July 7 confirming that Chinese AI models now account for between 30% and 46% of enterprise API token usage flowing through US developer platforms. Through OpenRouter, Chinese model share has been above 30% of all gateway tokens every week since February 8, rising as high as 46%. The average across the prior 12 months was just 11%, and had been as low as 4.5% in the first half of 2025.
Z.ai’s GLM-5.2 saw the fastest adoption of any model tracked by Vercel in 2026: daily token volume grew approximately 27x and customer count grew approximately 80x in its first full week after launch. Justin Summerville at OpenRouter quantified the price advantage: open-source Chinese models are 60 to 90 percent cheaper than leading Anthropic and OpenAI models. The Fable 5 export control ban, which restricted Anthropic’s most capable models for 19 days, likely accelerated this shift by pushing developers to explore alternatives.
Samsung Posts Record Q2 — Chip Division Profits Exceed 40-Year Cumulative Total
Samsung Electronics reported preliminary Q2 2026 operating profit of 89.4 trillion won ($58.4 billion) on 171 trillion won in revenue on July 7 — a roughly 19-fold year-over-year jump and a record for any tech company. The Device Solutions chip division booked 53.7 trillion won of the company’s first-quarter operating profit. DS president Kim Yong-kwan told a July 3 town hall that 2026 chip profit will exceed the cumulative total the division has earned across roughly 40 years in the business.
The numbers rest on memory prices that have surged through 2026, with commodity DRAM up around 90% in Q1 and 50–60% in Q2, as memory makers tilted capacity toward high-bandwidth memory (HBM) for AI workloads. SK Hynix, meanwhile, filed to list ADRs on Nasdaq starting July 10, with proceeds earmarked for new AI chip facilities. Samsung and SK Hynix together have committed 800 trillion won to a national semiconductor ecosystem project as part of South Korea’s $880 billion 10-year AI and chip investment plan.