The Great AI Awakening
Are we about to witness another inflection point in Generative AI's evolution? Cheaper reasoning models will proliferate. Here's why:
While we await GPT-5, Grok 4, DeepSeek-R2 and other frontier LLMs, there’s a lot to look forwards to in AI. I wanted to go over some of the AI reports I’ve been reading so far this summer. I expect you might not be back (completely yet) at your work desks yet so I’ll keep it light.
Will these new models upend the status-quo of LLM parity? I wanted to explore these reports in an eclectic mix of visuals meant to be skimmed.
🔥 The Big Three: Coming Soon ~ 𝙻𝚘𝚊𝚍𝚒𝚗𝚐
OpenAI’s GPT-5
xAI’s Grok 4
DeepSeek’s R2 reasoning model (was supposed go to live in May, 2025)
Explore my 2025 Site Map
List of Recent Articles
DeepSeek reasoning models are cheaper
When DeepSeek comes out with R2 we’ll get a better idea of the cost to performance of the best open-source reasoning models.
GPT-5 by OpenAI is expected to unify its systems with long-term memory and more personalization features.
Musk said in an X post that he and the xAI team were "grinding on Grok all night" with "good progress” on Grok 4 that is expected to be released in the weeks ahead.
Recently Iconiq Capital also released their own spin on the State of AI. The state of Building AI report.
The 2025 State of AI report, titled The Builder’s Playbook, provides a practical roadmap for AI innovation, focusing on the execution and scaling of AI-powered products. Drawing from an April 2025 survey of 300 software executives and interviews with AI leaders, the report highlights the strategies, challenges, and trends shaping the AI landscape.
If you know someone who enjoys both AI and infographics, feel free to share the piece:
This exploration looks at both the B2C (Menlo Ventures report) and B2B side (ICONIQ).
Of course I’m also going to share other tidbits and infographics I’ve found along the way.
Is Anthropic founding a “new species” of self-improving AI?
Claude Code just revealed that it's used by 115k developers and has changed 195M lines of code last week.
Claude Code was launched about 4 months ago, and you could argue Claude Code has rewritten more lines of code than most engineers see in a lifetime, and faster than most startups can file an expense report.
New grad hiring in BigTech is down -25% and -11% for startups
What does this brave new world look like exactly? 🤔
What type of AI products are you building?
Keep reading with a 7-day free trial
Subscribe to AI Supremacy to keep reading this post and get 7 days of free access to the full post archives.