We're happy to announce that Grok 3 is now available in GPTunneL — the newest large language model from Elon Musk's xAI. It's a powerful tool that opens up new possibilities for solving complex tasks and analyzing information in depth.
What makes Grok 3 special?
- The model excels at tasks that require deep analysis and logical reasoning. It doesn't just follow instructions — it understands complex, multi-step queries, breaking them into subtasks and working through them sequentially.
- The developers also significantly expanded the model's context window — it can now handle up to 128K tokens in a single chat.
- Grok 3 delivers excellent results in math and programming. The model can generate code, find bugs, and propose solutions for a variety of technical tasks, making it a valuable assistant for developers.
- Another key feature is its ability to work with real-time information thanks to web search access. This means answers based on the most current data, not just pre-trained knowledge.
Benchmark results
Source: Official Grok 3 announcement
Grok 3 shows strong results on standard language model benchmarks, outperforming many competitors. Here are a few examples from the model's official announcement:
- MMLU-Pro (broad language understanding across tasks): 79.9%
- GPQA (PhD-level problem solving): 75.4%
- AIME'24 (American math competition): 52%.
These numbers confirm the model's strengths in logic, math, and coding.
How to try it?
You can test the capabilities of the new Grok 3 model right now in GPTunneL
And one more piece of news: with the release of the new version, we've lowered the price on Grok 2. Access to the previous generation model is now even more affordable. Check out model pricing on our pricing page.