Gemini 3 and GPT 5.1
Caleb Writes Code
68,593 views • 8 months ago Save 3 min 2 min read
Video Summary
The latest AI model releases, including OpenAI's GPT 5.1 and Google's Gemini 3, highlight a significant acceleration in the AI industry's release cycles and a burgeoning competition between major players. Notably, Gemini 3's development entirely on Google's TPUs signals a potential shift away from reliance on Nvidia's hardware, a move that could have profound implications for the industry's infrastructure.
The relentless pace of AI innovation is underscored by rapid release cycles and the strategic timing of new model launches, with companies like Grok also contributing to the heated competition. As these advanced models emerge, the industry inches closer to Artificial General Intelligence (AGI), prompting discussions about automating AI development tasks and the critical importance of model credibility over mere performance metrics.
Short Highlights
Major AI Model Releases and Release Cycles [00:00]
- OpenAI's GPT 5.1, released November 12th, follows GPT5 released August 7th, marking a 97-day release cycle.
- Google's Gemini 3, released November 18th, follows Gemini 2.5 released March 25th, resulting in a 238-day cycle for major releases.
- Chinese open models such as Kim K2 (November 6th), Ling (October 9th), Miniax M2 (October 27th), and Quen 3 (April 28th) have demonstrated significant progress.
- American counterparts like Cloud 4, Cloud Opus 4.1, and Clad Son 4.5 are also noted for their impressive release cycles.
- Grok's last release was Grok 4 on July 9th, with a recent 4.1 upgrade, and anticipation for its next model is high.
"And we're also expecting big things from Grok since their last release of Gro 4 on July 9th with recent 4.1 upgrade and now we are anticipating what the next model holds."
Key Details
Industry Shifts and Hardware Independence [01:08]
- The AI industry has recently shifted its focus from Chinese models to American models, intensifying competition.
- Gemini 3 was trained exclusively on Google's TPUs, indicating a potential decrease in the AI industry's dependence on Nvidia's GPUs.
- Google is also lending its TPUs to companies like Anthropic and Midjourney, further decentralizing hardware reliance.
- Nvidia's earnings call demonstrated strong performance, beating analyst expectations for Q3 revenue and earnings per share.
"Now, that's not to say Nvidia isn't dire streets. Nvidia is doing just fine. In fact, their earnings call today show that they beat analyst expectation, not only in their Q3 revenue, but also in their earnings per share."
