xAI: Grok Imagine Video

This model is hosted by one provider. OpenRouter forwards every request to it directly — no routing decisions to make.

Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).

Scores on standardized evaluations. Higher percentages are better — and rank percentile shows where this model lands among all models on OpenRouter.

Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.

Token volume and request traffic to this model over time.

Percent of requests that succeeded over the last 30 days. OpenRouter monitors every provider continuously and automatically retries on the next-best provider when one returns an error.

Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.