:kotlin-intensifies-purple::robot_face: Curious ab...

# announcements

[Alina Dolgikh [JB]]

K🤖 Curious about how AI models handle Kotlin?

We put DeepSeek-R1, several OpenAI models, and others to the test using Kotlin-specific benchmarks. Here’s what we discovered:

✅ How well different models respond to Kotlin-related questions
✅ How the models compare in accuracy and reasoning
✅ Where they succeed – and where they struggle

👉 Read the full analysis: https://kotl.in/ai-models

[eygraber]

Why wasn't Grok included?

I used ChatGpt, Claude, Gemini, and Grok concurrently for a few months, and Grok was consistently the best, especially with Kotlin related prompts.

I still check in on the others every once in a while to see if there are any improvements, but I've switched to using Grok most of the time.

[Vera]

Thank you for sharing your opinion! On Kotlin_QA, Grok does get a pretty good 8.43/10. But on KotlinHumanEval, it's only 70.19%. It crashes 14.91% on tests, and 11.8% on compilation, most often on deprecated toLowerCase()/toUpperCase() and forgotten imports.

[eygraber]

Well it looks like Grok 3 (released today) fixes the toLowerCase/toUpperCase issue 😅

[darkmoon_uk]

I'd sooner see Mistral's models given attention than Grok, but that's perhaps more about the association than the models themselves. 🇪🇺 💪

The recent release of their claimed state-of-the-art Codestral model would make this interesting.