@@ -1,5 +1,5 @@
11---
2-summary: "Generate videos via video_generate from text, image, or video references across 14 provider backends"
2+summary: "Generate videos via video_generate from text, image, or video references across 16 provider backends"
33read_when:
44 - Generating videos via the agent
55 - Configuring video-generation providers and models
@@ -9,7 +9,7 @@ sidebarTitle: "Video generation"
99---
10101111OpenClaw agents can generate videos from text prompts, reference images, or
12-existing videos. Fifteen provider backends are supported, each with
12+existing videos. Sixteen provider backends are supported, each with
1313different model options, input modes, and feature sets. The agent picks the
1414right provider automatically based on your configuration and available API
1515keys.
@@ -116,6 +116,7 @@ generation.
116116| Google | `veo-3.1-fast-generate-preview` | ✓ | 1 image | 1 video | `GEMINI_API_KEY` |
117117| MiniMax | `MiniMax-Hailuo-2.3` | ✓ | 1 image | — | `MINIMAX_API_KEY` or MiniMax OAuth |
118118| OpenAI | `sora-2` | ✓ | 1 image | 1 video | `OPENAI_API_KEY` |
119+| OpenRouter | `google/veo-3.1-fast` | ✓ | Up to 4 images (first/last frame or references) | — | `OPENROUTER_API_KEY` |
119120| Qwen | `wan2.6-t2v` | ✓ | Yes (remote URL) | Yes (remote URL) | `QWEN_API_KEY` |
120121| Runway | `gen4.5` | ✓ | 1 image | 1 video | `RUNWAYML_API_SECRET` |
121122| Together | `Wan-AI/Wan2.2-T2V-A14B` | ✓ | 1 image | — | `TOGETHER_API_KEY` |
@@ -133,21 +134,22 @@ runtime modes at runtime.
133134The explicit mode contract used by `video_generate`, contract tests, and
134135the shared live sweep:
135136136-| Provider | `generate` | `imageToVideo` | `videoToVideo` | Shared live lanes today |
137-| --------- | :--------: | :------------: | :------------: | ---------------------------------------------------------------------------------------------------------------------------------------- |
138-| Alibaba | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` skipped because this provider needs remote `http(s)` video URLs |
139-| BytePlus | ✓ | ✓ | — | `generate`, `imageToVideo` |
140-| ComfyUI | ✓ | ✓ | — | Not in the shared sweep; workflow-specific coverage lives with Comfy tests |
141-| DeepInfra | ✓ | — | — | `generate`; native DeepInfra video schemas are text-to-video in the bundled contract |
142-| fal | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` only when using Seedance reference-to-video |
143-| Google | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; shared `videoToVideo` skipped because the current buffer-backed Gemini/Veo sweep does not accept that input |
144-| MiniMax | ✓ | ✓ | — | `generate`, `imageToVideo` |
145-| OpenAI | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; shared `videoToVideo` skipped because this org/input path currently needs provider-side inpaint/remix access |
146-| Qwen | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` skipped because this provider needs remote `http(s)` video URLs |
147-| Runway | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` runs only when the selected model is `runway/gen4_aleph` |
148-| Together | ✓ | ✓ | — | `generate`, `imageToVideo` |
149-| Vydra | ✓ | ✓ | — | `generate`; shared `imageToVideo` skipped because bundled `veo3` is text-only and bundled `kling` requires a remote image URL |
150-| xAI | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` skipped because this provider currently needs a remote MP4 URL |
137+| Provider | `generate` | `imageToVideo` | `videoToVideo` | Shared live lanes today |
138+| ---------- | :--------: | :------------: | :------------: | ---------------------------------------------------------------------------------------------------------------------------------------- |
139+| Alibaba | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` skipped because this provider needs remote `http(s)` video URLs |
140+| BytePlus | ✓ | ✓ | — | `generate`, `imageToVideo` |
141+| ComfyUI | ✓ | ✓ | — | Not in the shared sweep; workflow-specific coverage lives with Comfy tests |
142+| DeepInfra | ✓ | — | — | `generate`; native DeepInfra video schemas are text-to-video in the bundled contract |
143+| fal | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` only when using Seedance reference-to-video |
144+| Google | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; shared `videoToVideo` skipped because the current buffer-backed Gemini/Veo sweep does not accept that input |
145+| MiniMax | ✓ | ✓ | — | `generate`, `imageToVideo` |
146+| OpenAI | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; shared `videoToVideo` skipped because this org/input path currently needs provider-side inpaint/remix access |
147+| OpenRouter | ✓ | ✓ | — | `generate`, `imageToVideo` |
148+| Qwen | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` skipped because this provider needs remote `http(s)` video URLs |
149+| Runway | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` runs only when the selected model is `runway/gen4_aleph` |
150+| Together | ✓ | ✓ | — | `generate`, `imageToVideo` |
151+| Vydra | ✓ | ✓ | — | `generate`; shared `imageToVideo` skipped because bundled `veo3` is text-only and bundled `kling` requires a remote image URL |
152+| xAI | ✓ | ✓ | ✓ | `generate`, `imageToVideo`; `videoToVideo` skipped because this provider currently needs a remote MP4 URL |
151153152154## Tool parameters
153155@@ -389,6 +391,13 @@ only the explicit `model`, `primary`, and `fallbacks` entries.
389391(`aspectRatio`, `resolution`, `audio`, `watermark`) are ignored with
390392a warning.
391393</Accordion>
394+<Accordion title="OpenRouter">
395+Uses OpenRouter's asynchronous `/videos` API. OpenClaw submits the
396+job, polls `polling_url`, and downloads either `unsigned_urls` or the
397+documented job content endpoint. The bundled `google/veo-3.1-fast` default
398+advertises 4/6/8 second durations, `720P`/`1080P` resolutions, and
399+`16:9`/`9:16` aspect ratios.
400+</Accordion>
392401<Accordion title="Qwen">
393402Same DashScope backend as Alibaba. Reference inputs must be remote
394403`http(s)` URLs; local files are rejected upfront.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。