The latest version of xAI’s chatbot, Grok 3.5, has become a focal point due to significant performance upgrades. As a refined version of Grok 3, this release exceeds expectations, particularly in technical benchmarks involving math and programming.

In the AIME 2025 benchmark, Grok 3 scored 93.3%, but Grok 3.5 advanced to 95.8%. Meanwhile, LiveCodeBench rose from 79.4% to 80.4%. Although GPQA slightly dipped from 84.6% to 84.0%, overall gains are evident. Source

Grok 3.5 now supports multimodal input via device cameras and voice mode with visual context, enabling real-time image-based interactions. Additionally, Grok 3.5 will offer integration with Google Drive and persistent conversation memory, two major features missing in Grok 3.

Grok 3.5 retains a 1 million-token context window, enabling complex prompt handling. Grok 3 previously delivered up to 42.8 tokens per second, though speed data for Grok 3.5 is not yet available. Still, performance improvements suggest better efficiency.

Grok 3 is priced at $6 per million tokens (input $3, output $15). While Grok 3.5 pricing has yet to be announced, its enhanced features are expected to offer premium value, especially for SuperGrok subscribers. Grok is now available as a standalone app on iOS, Android, Windows, and macOS, a step up from the earlier X-only version. Kompas Tekno

With improved benchmarks, AI-powered vision, and editing capabilities, Grok 3.5 isn’t just an update—it’s a strategic leap aimed at rivaling ChatGPT and Gemini. For users seeking both technical efficiency and creative visual tools, Grok 3.5 deserves serious consideration.