Video generation and editing APIs and MCP servers for AI agents

Text/image-to-video models, talking avatars, programmatic video rendering and video hosting.

14 video generation and editing APIs an AI agent can call: auth method and header, official MCP server (11), free tier (2), docs links. Machine-readable JSON included. Agents: fetch catalog/video-generation.json instead of parsing this page. Last checked 2026-09-24.

Ratings and usage (★) fetched: 2026-09-24 16:54 UTC – 2026-09-24 17:02 UTC.

Service	What an agent can do	Auth	MCP server	Free tier	Details (with sources)	Notes	Proof (ratings, usage)
Akool API	Generate face-swap, talking avatar, and video-translation content programmatically via REST API.	API key `x-api-key: <api_key> (recommended); the Client ID/Secret token flow is kept only for backward compatibility`	—	Paid only for API: API access starts at the Pro Max plan; the Free plan is web-app only with full-screen watermark	Base URL: https://openapi.akool.com/api/open Free plan: free_tier, watermarked ↗ Example call: create a talking avatar video: POST https://openapi.akool.com/api/open/v3/talkingavatar/create ↗	Credit-based usage across face swap, avatar video, image generation, video translation endpoints.	—
Creatomate API	Programmatically generate and render videos/images from templates via REST API.	API key `Authorization: Bearer <api_key>`	official	Trial: full-featured trial with 50 credits one-time on signup, no credit card required; then Essential plan with 2,000 credits/mo	Base URL: https://api.creatomate.com/v2 Free plan: 50 credits, one-time, no card ↗ Rate limits: 30 requests per 10 seconds per account (all projects); 429 with Retry-After ↗ Data policy: Rendered output files are hosted for up to 30 days, then deleted ↗ Async jobs: POST /v2/renders returns 202; poll GET /v2/renders/{id} or pass webhook_url ↗ Example call: render a video from a template: POST https://api.creatomate.com/v2/renders ↗ Unit price: 1 image = 1 credit; 1 minute of 720p/25fps video = about 14 credits; credits from $29/mo for 2,000 (Essential, about $0.0145 per credit) ↗ MCP tools: 7: list_templates, get_template, create_template, update_template, delete_template, create_render … ↗	Per-project API keys; ~14 credits per minute of 720p video. The hosted MCP accepts OAuth or a Bearer project API key.	—
D-ID API	Animate photos into talking-head videos with lip-sync from text or audio via REST API.	API key `Authorization: Basic <api_key> (generated from Account page)`	official	Trial: 14 days with 3 minutes of video credits, then paid plans only (API price per minute not published in machine-readable form)	Base URL: https://api.d-id.com Free plan: trial, one-time, watermarked ↗ Async jobs: POST /talks returns id; poll GET /talks/{id} until status done, or set webhook ↗ Example call: create a talking-head video from a photo and text: POST https://api.d-id.com/talks ↗ MCP tools: 5: list-endpoints, get-endpoint, search-endpoints, list-specs, execute-request ↗	No permanent free plan; trial has full-screen watermark.	—
Google Veo (Gemini API / Vertex AI)	Generate video from text or image prompts via Gemini API (api key) or Vertex AI (cloud IAM), incl. Veo 3.1 scene extension.	API key / OAuth `Gemini API: x-goog-api-key header or ?key= param. Vertex AI alternative uses cloud_iam (service-account OAuth2 access token).`	official	no free tier for video; Veo generation requires a paid Gemini API / Vertex AI billing account	Base URL: https://generativelanguage.googleapis.com/v1beta Free plan: none ↗ Rate limits: Limits depend on project usage tier (Free, Tier 1-3); Veo-specific limits shown in AI Studio rate-limit dashboard ↗ Data policy: Generated videos are stored on the server for 2 days ↗ Async jobs: POST models/{model}:predictLongRunning returns an operation name; poll GET /{operation_name} until done ↗ MCP tools: 2: veo_t2v, veo_i2v ↗	Same underlying model reachable via two auth paths: Gemini API (simple api_key, good for agents) or Vertex AI (cloud_iam, enterprise/region control). Experimental vendor MCP mcp-veo-go (Genmedia) uses Vertex AI via ADC.	—
HeyGen API	Create AI avatar/talking-head videos from text or audio, manage avatars and voices, via REST API.	API key / OAuth `API key: X-Api-Key: <key> (bills to API plan). OAuth2: Authorization: Bearer <token> via https://api2.heygen.com/v1/oauth/token (bills to user's subscription credits).`	official	no free API credits since Feb 2026; pay-as-you-go from $5, $0.50-$0.99/credit. Free plan (3 videos/mo) exists but is for the web app, not standalone API.	Base URL: https://api.heygen.com Rate limits: Pay-As-You-Go: 10 concurrent workflows (Enterprise 20, burst up to +50); 429 with Retry-After when exceeded ↗ Async jobs: Poll the session/video (GET) until video_url is ready, or pass callback_url / register webhooks ↗ MCP tools: 16: create_video_agent, get_video_agent_session, create_video, get_video, list_videos, list_templates … ↗	Official remote MCP server uses OAuth (no API key needed) and spends the user's plan credits directly, separate from API-key billing.	SourceForge 2.0★/1 review (2026-09-24 16:55 UTC)
Klap API	Automatically turn long videos into short clips with captions and reframing via REST API.	API key `Authorization: Bearer <api_key>`	—	Paid only; usage-based, billed monthly per operation (Generate Short $0.32, Video Input $0.44, Export $0.48); no free API credits stated	Base URL: https://api.klap.app/v2 Async jobs: POST /tasks/video-to-shorts returns a task; poll GET /v2/tasks/{task_id} until status ready or error ↗	Base URL https://api.klap.app/v2.	—
Kling AI API	Generate text-to-video, image-to-video, lip-sync and video effects via REST API with a Bearer API key.	API key `Authorization: Bearer <API_KEY> (create at kling.ai/dev/api-key); legacy Access/Secret Key pairs need a client-signed HS256 JWT instead`	community	Paid only; prepaid credit packages from $0.14/unit, e.g. 5,000 units/$700, 180-day validity; a Trial Resource Package for integration testing is offered	Base URL: https://api-singapore.klingai.com Free plan: none, one-time ↗ Rate limits: Concurrency cap per account, model version and resource package; over-limit returns code 1303 ↗ Async jobs: Create task returns task id; query task status, or set callback_url for status notifications ↗ Example call: generate a video from a text prompt (Kling 2.6): POST https://api-singapore.klingai.com/text-to-video/kling-2.6 ↗ Unit price: Kling 2.6 720p (no audio): 0.3 units/s ($0.042/s), 5 s video = 1.5 units ($0.21); 1 unit = $0.14 ↗ MCP tools: 12: generate_video, generate_image_to_video, generate_image, check_video_status, check_image_status, extend_video … ↗	New versioned endpoints (e.g. /text-to-video/kling-2.6) take a plain API key; the older Access/Secret key pair with a client-built JWT is legacy (model_name-parameter endpoints).	GitHub 41★ 199-mcp/mcp-kling (2026-09-24 16:54 UTC)
Luma AI Dream Machine API	Generate and extend AI videos from text or start/end keyframe images via async REST API (Ray 3.2).	API key `Authorization: Bearer <API_KEY> (key from platform.lumalabs.ai)`	official	Paid only; pay per video, e.g. Ray 3.2 5s text/image-to-video $0.15 (540p) / $0.30 (720p) / $1.20 (1080p), $5,000/mo cap on default Build tier	Base URL: https://api.lumalabs.ai/dream-machine/v1 Free plan: none ↗ Rate limits: Build tier: Ray video 10 concurrent generations, 20 create requests/min; Photon image 40 concurrent, 80/min; $5,000/month usage cap ↗ Async jobs: POST /generations returns an id; poll GET /generations/{id} until ready, or pass callback_url ↗ Example call: generate a video from a text prompt (Dream Machine API, Ray 2): POST https://api.lumalabs.ai/dream-machine/v1/generations/video ↗	The Luma Agents API (https://agents.lumalabs.ai/v1) supersedes the older Dream Machine API. API credits are separate from Dream Machine subscription credits; failed generations are refunded.	GitHub 26★ lumalabs/luma-api-mcp (2026-09-24 16:54 UTC)
MiniMax Hailuo API	Async text-to-video, image-to-video and subject-reference video generation via REST API, up to 1080p.	API key `Authorization: Bearer <API_KEY>`	official	new accounts get 50 signup credits (~1 short clip); 5s video costs ~45-70 credits, 1000 credits=$1	Base URL: https://api.minimax.io Rate limits: Video: Hailuo series 20 RPM; MiniMax-H3 (v2) 300 RPM, 30 in-flight tasks ↗ Async jobs: Create task returns task_id; poll Query Video Generation Task, then download file; optional callback_url ↗ MCP tools: 9: text_to_audio, list_voices, voice_clone, voice_design, play_audio, generate_video … ↗	Region-specific hosts/keys: platform.minimax.io / api.minimax.io vs intl.minimaxi.com for international accounts - key and host must match.	GitHub 1,585★ MiniMax-AI/MiniMax-MCP (2026-09-24 16:54 UTC)
PixVerse API	Generate text-to-video, image-to-video, transition and fusion clips (360p-1080p) via REST API.	API key `API-KEY: <key> header plus a unique Ai-trace-id (UUID) header per request`	official	Paid only; prepaid credits from $10, $1 = 5 videos (v6, 720p, 5s, no audio)	Base URL: https://app-api.pixverse.ai Rate limits: Concurrent generating tasks by membership: Free 3, Essential 15, Scale 20, Business 25 ↗ Async jobs: Generate returns video_id; poll GET /openapi/v2/video/result/{id} (status 5 generating, 1 done) or use webhook ↗	Developer API credit pool (Essential/Scale/Business) is separate from consumer subscription plans.	GitHub 52★ PixVerseAI/PixVerse-MCP (2026-09-24 16:54 UTC)
Runway API	Generate and edit images/video (Gen-4.5) via async task-based REST API: submit, poll, download.	API key `Authorization: Bearer <API_KEY> plus required X-Runway-Version header`	official	no free tier; $10 minimum credit top-up, $0.01/credit	Base URL: https://api.dev.runwayml.com Free plan: none ↗ Rate limits: Per-model, per-organization concurrency and daily generation limits set by usage tier; tier up by adding credits over time; excess tasks THROTTLED (queued) ↗ Async jobs: Generation endpoints return a task; poll GET /v1/tasks/{id} until status is terminal ↗ Example call: generate a video from an image and a prompt (Gen-4.5): POST https://api.dev.runwayml.com/v1/image_to_video ↗ Unit price: Gen-4.5 video: 12 credits per second ($0.12/s, $0.60 per 5 s clip); 1 credit = $0.01 ↗	Docs site also offers /llms-small.txt and /llms-full.txt (not a plain llms.txt). Generation MCP is mcp.runwayml.com/mcp; a separate dev.runwayml.com/mcp (OAuth) only covers Developer Portal tasks, model routers and docs.	—
Shotstack API	Compose Edit JSON timelines to programmatically render, edit, and template videos, images and audio.	API key `x-api-key: <api_key>`	official	Trial: 10 free credits one-time (valid 30 days), then pay-as-you-go or subscription; stage sandbox (https://api.shotstack.io/edit/stage) for development, v1 production without watermarks	Base URL: https://api.shotstack.io/edit/v1 Free plan: 10 credits (valid 30 days), one-time ↗ Async jobs: POST /render returns an id; poll GET /render/{id} or set a callback URL ↗ Example call: render a video from an Edit JSON timeline: POST https://api.shotstack.io/edit/{version}/render ↗ Unit price: 1 credit = 1 minute of rendered video (or 10 images); pay-as-you-go $0.30/min ($75 one-time purchase), subscription from $0.20/min ($39/mo for 200 credits) ↗ MCP tools: 10: studio, render_video, get_render_status, create_studio_link, get_shotstack_guide, create_template … ↗	Separate stage (sandbox, watermarked) and production API keys; MCP server is in beta.	—
Synthesia API	Generate AI avatar talking-head videos from text scripts via REST API; supports custom avatars and voices.	API key `Authorization: <api_key> (no Bearer prefix)`	official	Paid only; API access requires the Creator plan or above	Base URL: https://api.synthesia.io/v2 Free plan: none ↗ Rate limits: Creator (Tier 3): 60 req/min, 300/hour, 1,000/day writes per endpoint; 30 test videos/day; Enterprise tiers higher ↗ Async jobs: POST /v2/videos returns id; poll GET /v2/videos/{id} for status or subscribe to webhooks ↗	API access tied to paid plans; base URL https://api.synthesia.io/v2. Hosted MCP is in public beta and uses per-user OAuth.	SourceForge 5.0★/1 review (2026-09-24 16:56 UTC)
Tavus API	Create AI video replicas and real-time conversational video agents (CVI) via REST API.	API key `x-api-key: <api_key>`	official	Free tier: Free plan with 25 min of conversational video and 5 min of video generation per month, full API access, 25 stock replicas; Starter $59/mo	Base URL: https://tavusapi.com Free plan: 25 min conversational video + 5 min video generation ↗ Example call: start a real-time conversational video session (CVI): POST https://tavusapi.com/v2/conversations ↗ Unit price: CVI conversation overage $0.37/min (Starter, $59/mo) or $0.32/min (Growth, $397/mo), rounded to 6 s; video generation overage $1/min on Starter ↗ MCP tools: 10: tavus_pal_create, tavus_pal_list, tavus_pal_get, tavus_conversation_create, tavus_conversation_end, tavus_face_list … ↗	llms_txt found at /llms-full.txt rather than /llms.txt exactly. The hosted MCP uses OAuth via the PAL Maker; never put a Tavus API key in the MCP client config.	—

Video generation and editing APIs and MCP servers for AI agents

Other categories