Grok Imagine Video 1.5 launches API access and 720p renders in 25 seconds

• xAI says xAI's API announcement puts Grok Imagine Video 1.5 into general availability via the API, while the linked official launch post frames it as an image-to-video upgrade focused on realism, physics, and speed.
• According to xAI's launch thread, the consumer-facing Video 1.5 Fast tier now renders 720p clips in about 25 seconds, down from 40-plus seconds on the previous model.
• xAI's trailer example shows the launch being pitched less as a benchmark drop and more as a creative-output play, with a full cinematic trailer made by @heavypulp.
• Early creator posts such as bennash's Grok short and hellorob's ComfyUI workflow already place Grok Imagine 1.5 inside multi-tool pipelines rather than as a one-click end point.

You can jump straight to the official launch post, open grok.com/imagine, and see xAI push two different surfaces at once: API access in the rollout post and a faster consumer mode in the main thread. The demo trailer in xAI's example post leans hard into glossy narrative footage, while one creator workflow shows Grok being slotted between ComfyUI, LTX, DaVinci, and Suno.

xAI's clearest concrete ship is API general availability. xAI's API announcement links to the launch post, and the main thread points creators to grok.com/imagine for the product surface.

That split matters because this is two rollouts in one sentence: the model is now an API product, and a faster variant is also being pushed to consumer users.

The speed claim is specific. According to xAI's launch thread, 720p renders now take about 25 seconds, versus 40-plus seconds on the previous model.

xAI pairs that latency cut with three quality claims in the same launch language: sharper realism, better physics, and faster generations, as stated in the thread and repeated in a community recap.

• Surface: consumer rollout, per xAI's API announcement
• Mode name: Video 1.5 Fast, per xAI's wording
• Resolution called out: 720p, per xAI's launch thread
• Claimed render time: about 25 seconds, per the thread
• Prior render time: 40-plus seconds, per the thread

For the showcase asset, xAI did not lead with a settings chart. xAI's trailer example points to a trailer by @heavypulp, which is a cleaner signal about the audience than the spec sheet is.

The clip positions Grok Imagine Video 1.5 as a narrative visual tool, not just a motion test bed for short loops.

The more interesting evidence in the tweet pool is what people are chaining around Grok. In hellorob's workflow post, Grok Imagine 1.5 sits inside a longer stack: Ideogram V4 for the Messi still, ComfyCloud MCP for other players, Grok for image-to-video, LTX 2.3 plus a FLW LoRA for transitions, DaVinci for speed ramping, a TimeSlice node, and Suno for audio.

That post turns the release into a practical pattern:

1. Generate a key image.
2. Expand the scene with MCP-connected tools.
3. Use Grok Imagine 1.5 for the image-to-video step.
4. Handle transitions in a separate video model stack.
5. Finish pacing and sound in edit tools.

A second creator pattern is mood-heavy shortform. bennash's Grok short and another bennash clip both credit Grok alongside Rendergeist Pulse, which suggests creators are already pairing xAI's model with external finishing or style tooling instead of treating Grok as the entire pipeline.

Those examples add a different data point than xAI's house trailer: eerie square-format shorts, posted before the official announcement window closed, were already circulating as soon as the model landed.