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.
Summary
The newsletter announces Grok 4.5, the first model developed jointly by SpaceXAI and Cursor following Cursor's $60B acquisition. Grok 4.5 delivers strong benchmarks comparable to Claude Opus 4.8 and GPT 5.5, while running at 80 tokens per second with 4x efficiency gains. Pricing is significantly undercut at $2/$6 per million tokens versus $5/$25 for Opus 4.8, with temporary free usage in Cursor and Grok Build. Elon Musk positioned it as an Opus-class model that is faster, more efficient, and cheaper, with a larger model coming next month.
OpenAI released GPT-Live, a next-generation voice model featuring full-duplex architecture that processes inputs continuously rather than using turn-based conversation. This enables simultaneous listening, processing, and output, reducing awkward pauses and supporting real-time translation. Users preferred GPT-Live over Advanced Voice Mode in approximately 75% of comparisons, with expert science scores nearly doubling.
ByteDance launched Seedream 5.0 Pro, an image generation model claiming to understand design with powerful editing capabilities approaching frontier level. The model features improved text rendering, structure alignment, precision editing with layer separation, and native multilingual support across 10+ languages.
Additional developments include Cognition's SWE-1.7 coding model nearing GPT-5.5 performance at lower cost, Joshua Achiam's departure from OpenAI after nine years, Prime Intellect's $130M Series A funding, and Eren Bali's Monogram emerging with $40M seed funding for an iOS app generating interfaces from text.
About this episode
PLUS: Add CGI to any video with Higgsfield
Key Insights
- SpaceXAI claims Grok 4.5 achieves 4x efficiency gains over models like Claude Opus 4.8 while maintaining comparable benchmark performance, positioning it as a value-driven alternative to existing frontier models.
- OpenAI argues that full-duplex architecture enabling simultaneous input processing and output generation creates more natural conversation experiences, with users preferring GPT-Live in 75% of head-to-head comparisons.
- ByteDance positions Seedream 5.0 Pro as transcending pure generation to provide design understanding through precision editing features with layer separation and multilingual capabilities.
- The industry narrative suggests that the gap between frontier models and cost-efficient alternatives is narrowing significantly, with multiple labs (SpaceXAI, Cognition) releasing competitive models at substantially lower prices.
- The shift in image and design models is moving from post-generation editing workflows toward more agentic, proactive systems that provide creative partnership and control within the initial generation process.
Topics
Transcript
Good morning, {{ first_name | AI enthusiasts }}. With its last few versions lagging behind, Cursor coming on board with its own models, and Elon Musk renting out compute to frontier rivals, Grok's place in the AI world seemed up in the air. The release of Grok 4.5, the first model SpaceXAI and Cursor trained together, just cleared things up fast — with powerful performance, flash speed, efficiency, and prices tuned for a moment when everyone is suddenly counting tokens. SpaceXAI and Cursor release the strongest Grok yet ChatGPT’s voice gets a major upgrade with GPT-Live Add CGI to any video with ChatGPT and Higgsfield ByteDance targets design work with Seedream 5.0 4 new AI tools, community workflows, and more…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Rundown AI
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.
Anthropic's Fable returns worldwide
Anthropic's Fable 5 AI model has returned to public access after a 3-week government-mandated shutdown, now featuring enhanced safety filters and requiring U.S. government pre-release access to future models. The newsletter also covers Meta's cloud compute rental plans, AI design tools, and new benchmarks showing Fable 5 achieving record performance on freelance task automation at 16.1% professional quality.