OpenAI sends GPT-5.6 to Work
OpenAI launched GPT-5.6 with three tiers (Sol, Terra, Luna), featuring near-Fable performance at lower costs and introducing ChatGPT Work with desktop integration. The newsletter also covers Meta's Muse Spark 1.1 release and argues that AI will concentrate demand among top 1-5% professionals while displacing mid-tier service providers.
Summary
OpenAI released the GPT-5.6 family, headlined by Sol, which achieves near-Fable performance on the Intelligence Index while excelling at agentic coding, computer use, and cybersecurity. Pricing remains competitive with GPT-5.5, ranging from $5/$30 per million tokens for Sol to $1/$6 for Luna. A notable feature is 'Ultra' mode for top performance, and Sol can autonomously post-train Luna. OpenAI launched ChatGPT Work as a Claude Cowork competitor, powered by Codex's engine and designed for accessible everyday tasks. The company is merging its Codex app into ChatGPT's revamped desktop application, which now includes built-in browser and computer control capabilities, representing OpenAI's move toward its 'Superapp' vision.
Rowan's Corner presents a compelling argument about AI's impact on service professions, drawing parallels to high-end portrait photography. The thesis contends that AI has become sufficiently good for standard applications, but premium customers will continue paying for top-tier professionals who can exceed AI capabilities by a meaningful margin. This creates a bifurcated market where the top 1-5% of professionals will capture concentrated demand while mid-tier service providers face displacement. The mechanism driving this shift is twofold: as AI substitutes for lower-tier services, the pool of paid human services shrinks and concentrates at the top, and AI-powered search recommendations accelerate this dynamic by directing users to single 'best' options rather than allowing browsing down rankings.
Meta released Muse Spark 1.1 with agentic reasoning, computer use capabilities, and a 1M-token context window, priced at $1.25/$4.25 per million tokens—approximately a quarter of competitor pricing. Zuckerberg claims the model is state-of-the-art or near-state-of-the-art on agent reasoning and tool use benchmarks. The newsletter draws parallels between Meta and SpaceX AI, noting both companies rebuilt failed initial efforts into surprisingly capable models with superior cost efficiency. Additionally, the newsletter includes a reader workflow showcasing how a commercial loan portfolio manager uses Claude to extract mortgage data, enrich contact information, and generate prioritized prospect lists with executive summaries.
About this episode
PLUS: Use 60% less Fable tokens with this orchestrator setup
Key Insights
- OpenAI positioned GPT-5.6 Sol as a near-Fable alternative that undercuts pricing and eliminates the usage constraints that create anxiety for Claude users on restrictive rate limits.
- Rowan argues that as AI becomes commoditized for standard work, premium service professionals will increasingly capture 10x more demand because AI-powered search and recommendations concentrate buyer attention on single top-rated providers rather than distributing it across ranked results.
- Meta's Muse Spark 1.1 achieved state-of-the-art or near-state-of-the-art agent reasoning performance while pricing at approximately one-quarter the cost of leading competitors, following Meta's pattern of rebuilding failed initial AI efforts into cost-efficient frontier models.
- The transcript presents a token-optimization strategy where high-capability models like Fable handle planning and final review while lower-cost workers handle routine, token-heavy tasks like coding and browsing, reducing overall spending.
- Rowan claims that mid-tier service professionals who cannot exceed AI capabilities by a meaningful margin will lose ground to the top 1-5%, as the total market for human services shrinks and remaining demand concentrates among premium providers.
Topics
Transcript
Good morning, {{ first_name | AI enthusiasts }}. OpenAI’s new 5.6 class is officially here, and while there's no Fable killer in the launch, there is a model that's nearly as good, far cheaper, and free of the usage tightrope Claude users walk with dread. GPT-5.6 Sol hits near-Fable scores at a fraction of the cost, paired with a new ChatGPT Work and desktop merge that brings the company’s superapp vision into users’ hands for the first time. OpenAI takes GPT-5.6 public, adds ChatGPT Work Rowan’s Corner: The top 1% economy Use 60% less Fable tokens with this setup Meta continues shipping with Muse Spark 1.1 4 new AI tools, community workflows, and more OPENAI The Rundown: OpenAI just released…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Rundown AI
SpaceXAI, Cursor release the strongest Grok yet
SpaceXAI and Cursor released Grok 4.5, a powerful new AI model that combines strong performance with faster speeds and lower costs than competitors like Claude and GPT. The newsletter also covers OpenAI's new GPT-Live voice model, ByteDance's Seedream 5.0 Pro image generation tool, and several other AI developments in the industry.
Meta climbs the AI image leaderboard
Meta released Muse Image, an in-house AI image model ranking No. 2 on leaderboards behind only OpenAI's GPT Image 2, with a teased video model also performing strongly. Meanwhile, Beijing is considering restrictions on Chinese AI model exports, creating potential geopolitical reciprocity risks similar to U.S. export controls, while companies like DoorDash demonstrate that pairing multiple AI models significantly improves code review accuracy and cost-effectiveness.
The part of Claude's brain nobody built
Anthropic researchers discovered 'J-space,' an internal workspace in Claude that wasn't explicitly programmed but emerged during training, functioning similarly to human conscious thought. The newsletter also covers Tencent's efficient Hy3 model, Kyutai's Rocket League world model, and practical AI applications including mobile app development with Replit.
Meta sizes up GPT-5.5 with 'Watermelon'
Meta's upcoming 'Watermelon' AI model reportedly matches OpenAI's GPT-5.5 performance while still in training, though the AI frontier continues advancing with competing models like Mythos and Fable. The newsletter covers Meta's $145B AI investment progress, practical AI applications from users, and emerging AI tools across education, coding, and consumer devices.
Altman invites Washington inside the AI industry
Sam Altman proposes a U.S.-led international AI safety forum modeled after the IAEA, with reports suggesting OpenAI discussed offering the government a 5% equity stake. Meanwhile, Meta's smart glasses capture 80% of the market but still fall short of replacing phones, and specialized AI models trained on expert data outperform frontier models at significantly lower costs.