Grok
xAI's chat model, accessed mainly through X; valued by shluchim as the model that refuses legitimate prompts least often, and for a less corporate writing tone.
Grok is xAI's chat model. Its distinct role in the group: the model that refuses least often on legitimate prompts other models block.
What it is
xAI's chat product, accessed primarily through X (formerly Twitter). A voice mode exists, and a Tesla maps integration was mentioned. It offers a generous free tier through X.
What shluchim use it for
- Writing with personality. Preferred over ChatGPT for casual or fun letters, for its "genuine tone." See Letter Writing.
- Image generation. A Grok-then-Nano Banana workflow succeeded when ChatGPT refused.
- Photo editing and manipulation.
- Logo design. Better than ChatGPT for a first-pass logo, though still behind Nano Banana.
- Prompts other models refuse. One member counted it among "the more 'kosher' ones."
- Video generation. At least one member moved from Sora to Grok for educational videos. See Video Creation.
Strengths
- Less likely to refuse benign prompts that other models over-block, for example rendering people in Chassidic or modest attire at a community event. For the reliable technique, see Depicting your community in AI images.
- Personality and tone, feels less corporate than ChatGPT.
- Generous free access through X.
Weaknesses
- Weak at complex Google Sheets formulas. One member: "Grok terrible."
- Less coverage of Jewish-content edge cases than the big three.
- Inconsistent on long-form work.
Open questions
- Whether the free version can create a video from an image, partially yes, depending on tier.
Related
- ChatGPT
- Gemini
- Letter Writing
- Flyer Design
- Find it in the Toolbox.