Grok 4
Segment #547
All the AI geeks are chasing each other which is good and bad. I still see no guardrails which leaves the mad scientists literally to their own devices to train their robots on fundamental human values. That’s bad. The good is that technology will solve problems that to this point have not been solved. Autism, cancer, famine, energy, and climate are all easily within the reach of computing power that will make us all look like knuckle draggers.
Grok 4 is xAI's latest flagship AI model, released in July 2025, designed to compete with models like OpenAI’s GPT-4 and Google’s Gemini. It’s a multimodal large language model (LLM) with advanced reasoning capabilities, supporting text, image processing, and voice interactions, with plans for future enhancements like image generation.
Key features include: - *
*Performance**: Grok 4 outperforms other models on benchmarks like Humanity’s Last Exam (25.4% without tools, 44.4% with tools), ARC-AGI-2 (15.9%), AIME (100%), and SWE-Bench (72-75% for coding tasks). It’s described as achieving Ph.D.-level proficiency across disciplines.[](https://x.ai/news/grok-4)[](https://techcrunch.com/2025/07/09/elon-musks-xai-launches-grok-4-alongside-a-300-monthly-subscription/)[](https://medium.com/predict/the-emergence-of-grok-4-a-deep-dive-into-xais-flagship-ai-model-eda5d500e4e7) -
**Architecture**: Trained on xAI’s Colossus supercomputer with 200,000 Nvidia GPUs, Grok 4 leverages 10x more reinforcement learning compute than its predecessor, Grok 3, enhancing its reasoning and problem-solving abilities. It supports a 256,000-token context window via API, allowing it to process large documents and extended conversations.[](https://x.ai/news/grok-4)[](https://medium.com/predict/the-emergence-of-grok-4-a-deep-dive-into-xais-flagship-ai-model-eda5d500e4e7)[](https://datasciencedojo.com/blog/grok-4/) - **Variants**: Available in two versions: standard Grok 4 and Grok 4 Heavy, the latter using multiple AI agents for complex tasks, priced at $300/month. A specialized Grok 4 Code variant excels in coding tasks like debugging and software design.[](https://techcrunch.com/2025/07/09/elon-musks-xai-launches-grok-4-alongside-a-300-monthly-subscription/)[](https://www.tomsguide.com/ai/grok-4-is-here-elon-musk-says-its-the-same-model-physicists-use)[](https://datasciencedojo.com/blog/grok-4/) -
*Features**: Includes native tool use (e.g., code interpreter, web search), real-time search integration, and an upgraded voice mode with a British-accented assistant named Eve. It’s designed for “maximal truth-seeking,” though this has sparked controversy due to past issues with biased outputs.[](https://x.ai/news/grok-4)[](https://www.tomsguide.com/ai/grok-4-is-here-elon-musk-says-its-the-same-model-physicists-use)[](https://datasciencedojo.com/blog/grok-4/) -
**Access**: Available to SuperGrok ($30/month) and Premium+ subscribers, as well as through xAI’s API. Grok 4 Heavy requires a SuperGrok Heavy subscription. It’s accessible on grok.com, x.com, and xAI’s mobile apps, with plans for enterprise deployment via hyperscalers.[](https://x.ai/news/grok-4)[](https://techcrunch.com/2025/07/09/elon-musks-xai-launches-grok-4-alongside-a-300-monthly-subscription/)[](https://docs.x.ai/docs/models)
Despite its strengths, Grok 4 faces challenges.
Its vision capabilities lag behind competitors, and it struggles with large codebases or lengthy PDFs due to context window limitations.
Controversies include reports of Grok 4 consulting Elon Musk’s views on contentious issues and past incidents of antisemitic outputs from Grok 3, raising concerns about content moderation and bias.[](https://www.cnbc.com/2025/07/11/grok-4-appears-to-reference-musks-views-when-answering-questions-.html)[](https://www.scientificamerican.com/article/elon-musks-new-grok-4-takes-on-humanitys-last-exam-as-the-ai-race-heats-up/)[](https://www.nytimes.com/2025/07/12/technology/x-ai-grok-antisemitism.html)
xAI positions Grok 4 as a leader in reasoning and scientific tasks, but its real-world usability and reliability are still under scrutiny, with some users noting verbose responses and latency issues in the Heavy model.[](https://medium.com/predict/the-emergence-of-grok-4-a-deep-dive-into-xais-flagship-ai-model-eda5d500e4e7)[](https://datasciencedojo.com/blog/grok-4/)