Wavespeed

WaveSpeedAI skill

Unified inference gateway. Async-by-default with polling. One API key for both media and LLM workloads.

When to invoke

• User asks to generate AI media (image / video / audio / avatar).
• User mentions a wavespeed-ai/* model uuid or vendor/model LLM id.
• User wants to upload a local file as input for image-to-X / video-to-X.
• User wants to list models or check WaveSpeed balance.
• User wants to call Claude / GPT / Gemini / Llama through the WaveSpeed LLM proxy.

Prereqs

export WAVESPEED_API_KEY="ws_..."     # https://wavespeed.ai/accesskey

CLI: ~/.local/bin/wavespeed-cli (uses ~/.local/share/wavespeed-skill/venv).

Subcommands

run — submit a media job, poll, return JSON

wavespeed-cli run wavespeed-ai/z-image/turbo '{"prompt":"Cat"}'
wavespeed-cli run wavespeed-ai/flux-dev '{"prompt":"...","output_format":"png"}' --timeout 600
wavespeed-cli run wavespeed-ai/seedance-v2 '{"image":"","prompt":"dolly in"}' --sync

Output URLs at .outputs[].

submit / result / cancel

wavespeed-cli submit wavespeed-ai/veo-3 '{"prompt":"..."}' --webhook-url https://app/cb
wavespeed-cli result  --wait
wavespeed-cli cancel 

upload — upload a local asset, print hosted URL

URL=$(wavespeed-cli upload ./input.png)
wavespeed-cli run wavespeed-ai/seedance-v2 \
  "$(jq -nc --arg u "$URL" '{image:$u, prompt:"camera pan"}')"

models / balance

wavespeed-cli models --names-only
wavespeed-cli models --filter video
wavespeed-cli balance

llm — OpenAI-compatible chat completion

wavespeed-cli llm anthropic/claude-opus-4.6 "Hello"
wavespeed-cli llm openai/gpt-5.2-pro "..." --system "Be terse."
wavespeed-cli llm google/gemini-3-flash-preview "..." --json-mode --raw
wavespeed-cli llm deepseek/deepseek-v4 "..." --stream

Base URL: https://llm.wavespeed.ai/v1.

verify-webhook

WAVESPEED_WEBHOOK_SECRET=ws_secret \
  wavespeed-cli verify-webhook "$RAW_BODY" "$X_WEBHOOK_SIGNATURE"

Python (scripting)

import wavespeed

out = wavespeed.run("wavespeed-ai/z-image/turbo", {"prompt": "Cat"})
print(out["outputs"][0])

from wavespeed import Client
client = Client(api_key="...", max_retries=3, max_connection_retries=5, retry_interval=1.0)
out = client.run(model, payload, timeout=600, poll_interval=1.0, enable_sync_mode=False)

url = wavespeed.upload("/path/to/image.png")

Run with the dedicated venv:

~/.local/share/wavespeed-skill/venv/bin/python myscript.py

Or ad-hoc:

uv run --with wavespeed python myscript.py

Frequently-used model uuids

| Family | uuid | |---|---| | Text→Image fast | wavespeed-ai/z-image/turbo | | Text→Image quality | wavespeed-ai/flux-dev, wavespeed-ai/flux-pro, wavespeed-ai/qwen-image-2 | | Text→Image w/ LoRA | wavespeed-ai/flux-dev-lora | | Image→Video | wavespeed-ai/seedance-v2, wavespeed-ai/kling-v2.1, wavespeed-ai/luma-dream-machine | | Text→Video | wavespeed-ai/veo-3, wavespeed-ai/veo-3-fast, wavespeed-ai/wan-2.7 | | Identity portrait | wavespeed-ai/higgsfield-soul-id | | Audio | wavespeed-ai/ace-step | | LLM | anthropic/claude-opus-4.6, openai/gpt-5.2-pro, google/gemini-3-flash-preview, deepseek/deepseek-v4, meta-llama/llama-4-70b |

Live catalog: wavespeed-cli models --names-only.

REST primitives (when SDK is overkill)

# Submit
curl -X POST "https://api.wavespeed.ai/api/v3/" \
  -H "Authorization: Bearer $WAVESPEED_API_KEY" -H "Content-Type: application/json" \
  -d '{...payload...}'

# Poll
curl "https://api.wavespeed.ai/api/v3/predictions//result" \
  -H "Authorization: Bearer $WAVESPEED_API_KEY"

Errors: 401 bad key, 429 rate limit, 400 payload mismatch, 402 insufficient balance.

Cost / latency knobs

• --timeout 600 image, --timeout 1800 video. Default 36000 is too long for interactive flows.
• --poll-interval 0.5 for fast image loops; default 1.0 is fine for video.
• --sync skips polling — only for sub-30s models.

Refs

• Docs: https://wavespeed.ai/docs
• SDK: https://github.com/WaveSpeedAI/wavespeed-python
• Models: https://wavespeed.ai/models
• API key: https://wavespeed.ai/accesskey
• Skill repo: https://github.com/wesleysimplicio/WaveSpeedAI-Skills

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: wesleysimplicio
• Source: wesleysimplicio/WaveSpeedAI-Skills
• License: MIT
• Homepage: https://wavespeed.ai

Install and usage instructions live in the source repository linked above.