GROK 4 Tested with code generation in Python Agents APP

🔍 Testing Grok-4 for Code Generation | MAIAChat Python Agents App Explore the performance of the newly released Grok-4, benchmarked alongside DeepSeek R1 70B and Gemini 2.5 Flash, with a focus on real-world code generation tasks. In this video, I test their capabilities using direct API calls via xAI and OpenRouter. While Grok-4 shows promise, current API response times are noticeably slow, and code generation can be time-consuming—areas likely to improve in the near future. The tests are performed using MAIAChat, a Python-based local agent framework I’ve developed and open-sourced. You can find the app here: 📂 GitHub Repository: https://github.com/alexcelewicz/MAIAChat-Desktop What’s Included: Real-time performance tests of Grok-4, DeepSeek R1 70B, and Gemini 2.5 Flash API call comparison and latency insights Code generation examples: website scaffolding and simple p5.js game creation Early findings with more detailed benchmarking to follow 📌 Stay tuned for a deeper dive into capabilities, stability, and use cases in upcoming videos.