Retter Reboot #28

The Retter Reboot : Your weekly dose of ai & tech news

1. OpenAI Ships GPT-5.6: Sol, Terra, and Luna Go Generally Available

OpenAI moved its GPT-5.6 family from restricted preview to general availability on July 9, rolling out globally across ChatGPT, Codex, and the API after coordination with the US government. Sol is the flagship for hard reasoning at $5/$30 per million tokens, Terra the balanced mid-tier at $2.50/$15, and Luna the fast, cheap option at $1/$6. OpenAI is pitching the family on capability per dollar, claiming fewer tokens, lower cost, and faster completion on agentic tasks. After a preview limited to about 20 trusted partners, the frontier lineup just got a full refresh.

Source: Coursiv

2. ChatGPT Work Launches: OpenAI's Agent That Ships Finished Deliverables

Alongside GPT-5.6, OpenAI unveiled ChatGPT Work on July 9, an agent that takes a goal, gathers context across your connected apps and files, and works independently for hours to return finished documents, spreadsheets, presentations, and even hosted websites. It has Codex technology built in, plus control surfaces like Plan mode and action approvals. Rollout starts with Pro, Enterprise, and Edu, expanding to Plus and Business within days. It is OpenAI's direct answer to Claude Cowork and Microsoft's Copilot Cowork, and it confirms agents-that-deliver-artifacts as the battleground for workplace AI.

Source: Reuters

3. Anthropic Extends Fable 5 on Paid Plans Through July 12

Hours before Claude Fable 5 was due to drop off paid subscriptions, Anthropic extended included access through July 12 at 11:59 PM PT, following customer pushback over the early cutoff. Pro, Max, Team, and eligible Enterprise subscribers can keep using the Mythos-class model for up to 50% of weekly limits at no extra cost, and Anthropic even reset everyone's weekly limits on July 9, conveniently timed with OpenAI's GPT-5.6 launch. After the deadline, Fable 5 moves to usage credits at $10/$50 per million tokens, though Anthropic says it aims to restore it to standard subscriptions once capacity allows. If you have quota left, this weekend is the time to burn it on your hardest problems.

Source: Android Authority

4. Microsoft Starts Swapping OpenAI and Anthropic for Its Own Models in Office

Bloomberg reported that Microsoft has begun routing tens of thousands of weekly Excel and Outlook Copilot prompts to its in-house MAI models instead of OpenAI and Anthropic, targeting high-volume routine tasks where cost favors a cheaper first-party model. AI chief Mustafa Suleyman was blunt about the motive, saying the goal is to reduce and ultimately eliminate what Microsoft pays Anthropic. The model behind a familiar Copilot button can quietly change, and workload-level routing is becoming the enterprise norm.

Source: Bloomberg

5. Chinese Open-Weight Models Now Take Up to 46% of US Enterprise API Traffic

A CNBC investigation found Chinese AI models now account for between 30% and 46% of enterprise API token usage on US developer platforms, up from roughly 11% on average over the prior year. Z.ai's GLM-5.2 is the standout, scoring competitively with Western frontier models on SWE-bench Pro at a fraction of the price and posting 80x customer growth in one week on Vercel. The emerging playbook: cheap open-weight model as default, frontier Western model reserved for the hardest tasks.

Source: Build Fast with AI

6. AI Infrastructure Swallows 79% of the Week's $9.9B in Venture Funding

Four of every five venture dollars this week went to AI infrastructure. Together AI closed an $800 million Series C led by Aramco Ventures after crossing $1.15 billion in annual bookings, with plans to grow its cloud capacity 50x over five years. Crusoe entered talks for $3 billion at a valuation near $18 billion, Switch is seeking $2 billion, and Schneider Electric paid $3.1 billion to acquire industrial data platform Cognite. The capital is chasing one conviction: whoever controls inference capacity controls the next phase of enterprise AI.

Source: StartupHub.ai

7. Salesforce Leads $135M Round for Agentic Software Development

8090 Solutions, co-founded by investor Chamath Palihapitiya, raised $135 million led by Salesforce to scale its Software Factory platform, a system where coordinated AI agents build enterprise software under human-led oversight, targeting regulated sectors like healthcare and aerospace. Founded only in 2024, the company is the clearest signal yet that investors believe agent teams, not individual copilots, are the future of enterprise software delivery. For anyone in custom software, this is the competitive model to watch.

Source: Crunchbase

8. xAI Launches Grok 4.5

xAI announced Grok 4.5, positioned around stronger coding, agentic task handling, and knowledge work. Benchmarks, pricing, and availability were not verifiable from the launch page at time of writing, so treat the capability claims as unconfirmed for now. Still, it caps a week in which every major lab shipped something, and the release cadence shows no sign of slowing.

Source: Future Tools

Jul 10, 2026