Eseandre Mordi
⬤ xAI is running experiments with its upcoming model iteration, with several Grok 4.1 versions now being tested on platforms like LmArena and OpenRouter. This testing phase suggests xAI is gearing up for a new release while continuing to fine-tune its current systems. However, expectations should remain measured—Grok 4.1 isn't expected to deliver a massive performance leap, with more substantial improvements likely reserved for Grok 5.
⬤ The current phase focuses on testing multiple Grok 4.1 variants across external platforms. By using sites like LmArena and OpenRouter, xAI can expose the new versions to diverse real-world prompts and usage scenarios, helping the team evaluate how these iterations perform in practice. The company is framing Grok 4.1 as an incremental step rather than a complete redesign—essentially a bridge to more ambitious upgrades down the line.
⬤ xAI appears to be gearing up for the Grok 4.1 release, with new model identifiers now visible in Grok.com's source code. The names include grok-4-1-non-thinking-w-tool and grok-4-1-non-thinking-no-tool-1111b, pointing to at least 2 different Grok 4.1 configurations in development. These references in the site's source suggest that a wider rollout is getting closer.
⬤ The combination of external platform testing and the appearance of Grok 4.1 model names in the source code shows that xAI is steadily advancing its Grok lineup. For those tracking AI model development, Grok 4.1 represents ongoing progress and preparation ahead of the more ambitious improvements expected with Grok 5.
Eseandre Mordi
