xAI ships Grok Imagine 1.5 (720p image-to-video) and rolls out Grok Voice; Grok Build 0.1 coding model enters API beta

xAI shipped a cluster of product releases in early June 2026. On June 3-4 it released Grok Imagine 1.5 Preview (model id grok-imagine-video-1.5-preview), an image-to-video model available via the xAI API that converts a single still image into cinematic video at up to 720p with natural-language motion control, preserving the source image's look and lighting. Per reporting, the model generates native synchronized audio — background music, sound effects and lip-synced dialogue — extends clips to 15 seconds, and debuted at #1 on the Artificial Analysis Video Arena image-to-video leaderboard with an Elo of 1404.

Alongside it, xAI publicly rolled out Grok Voice (spoken interaction with the model) on June 4. Separately, in late May (May 28-29) xAI put Grok Build 0.1 (grok-build-0.1) into public beta on its API — positioned as a fast agentic coding model running at 100+ tokens/second with a 256,000-token context window, always-on reasoning, text and image inputs, priced at $1 per million input tokens and $2 per million output tokens. The same May 29 developer drop added a Smart Turn streaming speech-to-text end-of-turn detector, a Context Compaction API to shrink conversations for lower cost, a WebSocket Responses API mode for tool-heavy agent workloads, and image search within web search.

Collectively the releases push xAI further into multimodal generation (video + native audio + voice) and agentic coding, narrowing the product surface gap with OpenAI, Google and Anthropic. The video and voice features are preview/beta, not general availability.