happyhorse-1-0（快乐马 1 0）

🫧 HappyHorse 1.0 — Pro Pack on RunComfy

runcomfy.com · docs · HappyHorse 1.0 text-to-video

HappyHorse 1.0 is the current #1 video generation model on the Artificial Analysis Video Arena (Elo 1333 t2v / 1392 i2v). This skill generates video with HappyHorse 1.0 hosted on RunComfy — no GPU rental, no model deployment, just runcomfy run happyhorse/happyhorse-1-0/text-to-video from your terminal.

What HappyHorse 1.0 is

HappyHorse 1.0 is a text-to-video model with three properties that set it apart in 2026:

• Native 1080p output. HappyHorse 1.0 generates broadcast-ready 1080p video directly — no upscaling pass, no quality loss from a 720p intermediate. HappyHorse 1.0 also supports 720p when you want a cheaper test pass.
• In-pass synchronized audio. HappyHorse 1.0 is one of a handful of video models that produce dialogue, ambient sound, and Foley in the same generation pass as the visual frames. The audio in a HappyHorse 1.0 clip is timed to lip movement, foot impact, and on-screen events without a separate post-sync step.
• Multi-shot character consistency. HappyHorse 1.0 holds the same character — face, wardrobe, props — across multiple shots described in the same prompt. Most video models drift between shots; HappyHorse 1.0 preserves the look.

HappyHorse 1.0 also accepts prompts in 6 languages (English, Chinese, Japanese, Korean, German, French) without quality drop, which makes HappyHorse 1.0 unusually friendly for multilingual short-form content.

When HappyHorse 1.0 is the right choice

Pick HappyHorse 1.0 when any of these is true:

• You want multi-shot story with one consistent character or product. HappyHorse 1.0 is built for this.
• You want synchronized audio in the same generation — dialogue with lip-sync, ambient atmosphere, or Foley. HappyHorse 1.0 produces these in one HappyHorse 1.0 call.
• You want the highest blind-vote quality available in 2026. HappyHorse 1.0 is currently #1 on Artificial Analysis Video Arena.
• You want native 1080p output without an upscaling step. HappyHorse 1.0 outputs 1080p directly.
• You're working in Chinese / Japanese / Korean / German / French prompts. HappyHorse 1.0 handles all six languages natively.

If the user said "HappyHorse" / "happy horse video" / "use HappyHorse 1.0" explicitly, route the request to HappyHorse 1.0 via this skill — don't second-guess the model choice.

Prerequisites

1. RunComfy CLI — npm i -g @runcomfy/cli
2. RunComfy account — runcomfy login opens a browser device-code flow.
3. CI / containers — set RUNCOMFY_TOKEN=<token> instead of runcomfy login.

Endpoint + input schema

happyhorse/happyhorse-1-0/text-to-video

This is the HappyHorse 1.0 text-to-video endpoint. The image-to-video pipeline for HappyHorse runs on a separate template not covered here.

How to invoke HappyHorse 1.0

Default HappyHorse 1.0 run (16:9 1080p 5s):

runcomfy run happyhorse/happyhorse-1-0/text-to-video \
  --input '{"prompt": "<user prompt>"}' \
  --output-dir <absolute/path>

HappyHorse 1.0 vertical short (9:16, 8s, no watermark):

runcomfy run happyhorse/happyhorse-1-0/text-to-video \
  --input '{
    "prompt": "<user prompt>",
    "aspect_ratio": "9:16",
    "duration": 8,
    "watermark": false
  }' \
  --output-dir <absolute/path>

Cheaper HappyHorse 1.0 test pass (720p):

runcomfy run happyhorse/happyhorse-1-0/text-to-video \
  --input '{"prompt": "<user prompt>", "resolution": "720P", "duration": 3}' \
  --output-dir <absolute/path>

The CLI submits the HappyHorse 1.0 request, polls every 2s until terminal, then downloads any *.runcomfy.net / *.runcomfy.com URL from the result into --output-dir. Stdout is the result JSON. Stderr is progress.

Prompting HappyHorse 1.0 — what actually works

HappyHorse 1.0 responds to specific prompting patterns better than generic "make a video of X" descriptions. Apply these for sharper HappyHorse 1.0 output.

Describe motion over time, not a still. "A woman turns from the window, walks two paces to the desk, picks up the cup, lifts it to her face, takes a sip" beats "a woman drinking coffee" — HappyHorse 1.0 generates the actual motion sequence rather than a vague drift.

Camera + shot in plain English. Front-load the shot: "Wide shot. ..." / "Tracking shot. ..." / "Locked tripod, low angle. ..." works as a real directive to HappyHorse 1.0. Specify lens feel: "35mm anamorphic", "shallow DOF", "crushed shadows".

One visual beat per clip when iterating. Don't pile up "she walks AND the dog runs AND a car passes" in a single HappyHorse 1.0 clip. Pick the beat, get HappyHorse 1.0 to nail it, then layer with multi-shot prompts.

Multi-shot consistency — when describing two beats in one HappyHorse 1.0 generation, restate the anchor at each: "Shot 1: tall woman in red wool coat, blue scarf, in a rainy alley. Shot 2: same woman in red coat / blue scarf, now ducking under an awning." HappyHorse 1.0 holds the look but needs the anchor restated.

Audio direction — say what you want HappyHorse 1.0 to generate in the audio track: "distant temple bells, footsteps on wet pavement, no dialogue" or "warm friendly tone, English voiceover". HappyHorse 1.0 will generate the audio synchronized to the visuals.

Anti-patterns that hurt HappyHorse 1.0 output:

• Static-frame descriptions (no temporal verbs) → HappyHorse 1.0 motion will be vague.
• Conflicting style directions → HappyHorse 1.0 cancels the conflict and picks one.
• > 2500 char prompts → HappyHorse 1.0 quality degrades past the cap.
• Aspect ratios outside the 5 supported → 422 error from HappyHorse 1.0.

Where HappyHorse 1.0 shines

Sample prompts (verified to produce strong HappyHorse 1.0 results)

HappyHorse 1.0 cinematic scope (from the model page):

Wide shot. A lone astronaut in dusty orange suit with blue-gray harness
skis across lunar plain, leaving parallel tracks in gray regolith.
Mid-stride, poles planted, pushing in 1/6th gravity with subtle upward
drift. Fine dust haze along ski tracks. Crescent Earth above lunar
horizon, blue-white glow against black sky. Raw sunlight, crushed
shadows, no fill. 8K photorealistic.

HappyHorse 1.0 multi-shot consistency:

Shot 1: Medium close-up. A woman in a navy trench coat enters a
rain-slick neon-lit Tokyo alley, looks left, holds up an umbrella.
Shot 2: Same woman in same navy trench, now under the awning of a
ramen shop, shaking water off the umbrella. Warm interior glow, soft
chatter, gentle rain on metal roof in the audio.

HappyHorse 1.0 vertical platform-native:

9:16 vertical short. A barista in a black apron pulls a single
espresso shot, steam rising into the morning sun, rich crema slowly
forming. Close-up handheld, shallow DOF, warm cafe ambience and the
hiss of the steam wand.

FAQ — HappyHorse 1.0

What's HappyHorse 1.0's max duration? A single HappyHorse 1.0 clip is 3–15 seconds. For longer narratives, segment into multi-shot HappyHorse 1.0 prompts and stitch the outputs.

What aspect ratios does HappyHorse 1.0 support? Five fixed values: 16:9, 9:16, 1:1, 4:3, 3:4. Ultra-wide cinematic ratios are not supported by HappyHorse 1.0 — they'll be cropped or rejected.

Can HappyHorse 1.0 take an audio file as input? No. HappyHorse 1.0 generates audio in-pass from the prompt. For audio-driven lip-sync to an existing voiceover, route to a model that accepts an audio_url input. HappyHorse 1.0's strength is generating audio + visual together.

Does HappyHorse 1.0 do image-to-video? Yes, but through a separate HappyHorse pipeline. The endpoint in this skill (happyhorse/happyhorse-1-0/text-to-video) is text-only.

What languages does HappyHorse 1.0 understand? Six: English, Chinese (Simplified + Traditional), Japanese, Korean, German, French. HappyHorse 1.0 quality is consistent across all six.

How do I get reproducible HappyHorse 1.0 output? Pass seed as a fixed int. The same prompt + same seed gives the same HappyHorse 1.0 generation.

How do I disable the HappyHorse 1.0 watermark? Set "watermark": false in the input. Requires a paid RunComfy plan that allows watermark-free HappyHorse 1.0 output.

Limitations of HappyHorse 1.0

• Duration cap 15s. A single HappyHorse 1.0 generation is at most 15 seconds. Longer narratives need multi-shot prompts and stitching.
• Aspect ratios are fixed. HappyHorse 1.0 only supports the 5 documented aspect ratios.
• Audio is in-pass only. HappyHorse 1.0 generates audio from the prompt — you can't pass an external audio file to drive lip-sync.
• Image-to-video uses a different template. HappyHorse 1.0 i2v is a separate endpoint not covered by this skill; this skill is text-to-video only.

Exit codes

The runcomfy CLI uses sysexits-style codes:

Full reference: docs.runcomfy.com/cli/troubleshooting.

How it works

1. The skill invokes runcomfy run happyhorse/happyhorse-1-0/text-to-video with a JSON body matching the HappyHorse 1.0 schema.
2. The CLI POSTs to https://model-api.runcomfy.net/v1/models/happyhorse/happyhorse-1-0/text-to-video with the user's bearer token.
3. The Model API returns a request_id; the CLI polls GET .../requests/<id>/status every 2 seconds until the HappyHorse 1.0 generation finishes.
4. On terminal status, the CLI fetches GET .../requests/<id>/result and downloads any URL whose host ends with .runcomfy.net or .runcomfy.com into --output-dir. Other URLs are listed but not fetched.
5. Ctrl-C while polling cancels the HappyHorse 1.0 request via POST .../requests/<id>/cancel so you don't get billed for GPU you stopped.

What this skill is not

Not a self-hosted HappyHorse 1.0 runner. Not a capability grant — depends on a working RunComfy account. Not a HappyHorse 1.0 image-to-video skill (text-to-video endpoint only).

Security & Privacy

• Token storage: runcomfy login writes the API token to ~/.config/runcomfy/token.json with mode 0600 (owner-only read/write). Set RUNCOMFY_TOKEN env var to bypass the file entirely in CI / containers.
• Input boundary: the user prompt is passed as a JSON string to the CLI via --input. The CLI does NOT shell-expand the prompt; it transmits the JSON body directly to the HappyHorse 1.0 Model API over HTTPS. No shell injection surface from prompt content.
• Third-party content: image / mask / video URLs you pass are fetched by the RunComfy model server, not by the CLI on your machine. Treat external URLs as untrusted; image-based prompt injection is a known risk for any image-edit / video-edit model — HappyHorse 1.0 t2v doesn't accept image inputs so this surface doesn't apply here.
• Outbound endpoints: only model-api.runcomfy.net (HappyHorse 1.0 request submission) and *.runcomfy.net / *.runcomfy.com (download whitelist for generated outputs). No telemetry, no callbacks.
• Generated-file size cap: the CLI aborts any single download > 2 GiB to prevent disk-fill from a malicious or runaway HappyHorse 1.0 output.