Keys & plans

API keys, models and plans

There are three ways to pay for the AI a video uses: your own API keys, our keys on the All-inclusive plan, or your own AI app over MCP on its own subscription. Plans differ only in how many minutes of video they include; every plan has the same Director and the same features.

Plans

Plan	Video minutes	Computers	Whose AI keys
Free	10 min a month	1	Yours (Gemini and any provider below)
Standard	60 min	1	Yours (Gemini and any provider below)
Pro	180 min	2	Yours (Gemini and any provider below)
Enterprise	600 min	10	Yours (Gemini and any provider below)
All-inclusive (our keys)	60 min	1	Ours — no keys needed

A video is charged for its real length, never more than the length it was quoted. A failed or cancelled video costs nothing.
"Computers" is how many computers can use the plan at once.
Prices are on the pricing page.

Whose AI pays for what

You use…	Who pays the AI provider	What you set up
Your own keys (Free, Standard, Pro, Enterprise)	You, on your provider account. Gemini's free AI Studio key is enough to start.	Add a key in Settings → API keys
Our keys (All-inclusive)	Included in the plan	Nothing
Your AI app over MCP	Your AI app's own subscription or sign-in (Claude, ChatGPT/Codex, OpenCode…)	Connect it. Voice and pictures still use the keys above.

Add a key

Get a key from the provider. For Gemini: Google AI Studio → Get API key (free).
In the app, open Settings → API keys, pick the provider and paste the key.
Press Test. The app checks the key with the provider and shows whether it works.

Settings, API keys: providers with their keys (hidden), health and a Test button — Settings → API keys. Key values are never shown in full.

As many keys as you like per provider. Calls rotate across them, and a key that is cooling down or out of today's quota is skipped.
Reserve a key for some stages, for example one Gemini key only for research.
Keys are stored encrypted on your computer. Full key values never leave the app, except to call their own provider.

Providers with their own entry

Entry on the API keys page	What it is used for
Google AI Studio (Gemini)	Deep Research, understanding, Director, relabel, sound director
Firecrawl	web search + scraping
Exa	neural search, image search
Serper (Google search API)	deep research web / news / scholar / video search
Deepgram	Deepgram voices + TTS fallback
Bria (background removal API)	cut-outs by the Bria API when the video's engine is bria_api
PhotoRoom (segmentation API)	cut-outs by PhotoRoom when the video's engine is photoroom_api
Fish Audio	Fish Audio voices (incl. your clones)
ElevenLabs	ElevenLabs voices (via LiteLLM)
OpenAI	OpenAI models + voices (via LiteLLM)
Anthropic (Claude)	Claude models for any usage you pick (via LiteLLM)
Proxy (your own)	YouTube downloads (and optional scraping) through your proxy — http, https, socks5
YouTube cookies	a cookies.txt export so YouTube downloads work when it asks to confirm you're not a bot
Google Vertex AI (your Google Cloud project)	pictures in batch at half price (and Vertex models) on YOUR project -- enter project-id or project-id/region; sign-in: your gcloud login (ADC) or a service-account key below
Google Vertex AI -- service-account key (optional)	paste the service-account JSON instead of using your gcloud login

Models: Auto, or your choice

Every AI role (script, director, vision, voice, pictures…) uses Auto by default: the best model your keys can reach, with a fallback when it is busy. To pick one yourself, open Settings → Models and choose a model for that role. You can also set how hard it thinks (reasoning: auto, off, low, medium, high) where the model supports it. The list of models is read live from your keys, so new models appear by themselves.

Settings, Models: each role with Auto or a chosen model — Settings → Models. Auto shows the model it would pick right now.

Every supported provider (via LiteLLM)

Beyond the entries above, the app works with every provider in the LiteLLM library built into it. Pick the provider on the API keys page and paste its key. Gemini is not in this list: it has its own, faster path in the app, including Gemini voices.

Key only means an API key is all it needs. Not yet means the provider also needs something the keys page cannot hold yet, such as an endpoint address, a region or a cloud account. The reason is shown next to it.

82 providers with a model list in LiteLLM 1.97.0 (the version inside the app). The numbers are how many models of each kind LiteLLM lists.

Provider	What it can do	Key variable	In the app
Fireworks AI `fireworks_ai`	text (321), pictures (9), transcripts (2), embeddings (3)	`FIREWORKS_AI_API_KEY`	Key only
Azure OpenAI `azure`	text (244), voice (5), pictures (33), transcripts (9), video (3), embeddings (12)	`AZURE_API_KEY`	Not yet needs your Azure endpoint and API version as well as the key
AWS Bedrock `bedrock`	text (260), pictures (18), embeddings (21)	`BEDROCK_API_KEY`	Not yet needs an AWS access key pair and a region
fal.ai `fal_ai`	text (1), pictures (201), video (8)	`FAL_AI_API_KEY`	Key only
OpenAI `openai`	text (125), voice (7), pictures (56), transcripts (9), video (3), embeddings (4)	`OPENAI_API_KEY`	Key only
Azure Ai `azure_ai`	text (121), pictures (10), transcripts (1), embeddings (5)	`AZURE_AI_API_KEY`	Not yet needs your Azure AI endpoint as well as the key
Deepinfra `deepinfra`	text (136)	`DEEPINFRA_API_KEY`	Key only
Novita `novita`	text (130), embeddings (3)	`NOVITA_API_KEY`	Key only
Vercel Ai Gateway `vercel_ai_gateway`	text (93), embeddings (6)	`VERCEL_AI_GATEWAY_API_KEY`	Key only
Together AI `together_ai`	text (90), embeddings (3)	`TOGETHER_AI_API_KEY`	Key only
Perplexity `perplexity`	text (84), embeddings (4)	`PERPLEXITY_API_KEY`	Key only
Databricks `databricks`	text (76), pictures (4), embeddings (3)	`DATABRICKS_API_KEY`	Not yet needs your Databricks workspace address
Mistral `mistral`	text (51), voice (4), transcripts (12), embeddings (4)	`MISTRAL_API_KEY`	Key only
Alibaba DashScope `dashscope`	text (43), voice (5), pictures (4), video (11)	`DASHSCOPE_API_KEY`	Key only
Nebius `nebius`	text (58), embeddings (4)	`NEBIUS_API_KEY`	Key only
xAI (Grok) `xai`	text (43), pictures (8), transcripts (2), video (5)	`XAI_API_KEY`	Key only
Bedrock Mantle `bedrock_mantle`	text (50)	`BEDROCK_MANTLE_API_KEY`	Not yet needs an AWS access key pair and a region
Github Copilot `github_copilot`	text (47), embeddings (3)	`GITHUB_COPILOT_API_KEY`	Not yet signs in with GitHub, not an API key
Oci `oci`	text (41), embeddings (9)	`OCI_API_KEY`	Not yet needs an Oracle Cloud tenancy, user, key file and region
Google Vertex AI `vertex_ai`	text (24), voice (10), pictures (1), transcripts (6), embeddings (2)	`VERTEX_AI_API_KEY`	Not yet needs a Google Cloud project and sign-in (the app's own Vertex AI entry covers pictures)
Deepgram `deepgram`	transcripts (42)	`DEEPGRAM_API_KEY`	Key only
Replicate `replicate`	text (40)	`REPLICATE_API_KEY`	Key only
Snowflake `snowflake`	text (35), embeddings (2)	`SNOWFLAKE_API_KEY`	Not yet needs your Snowflake account and a signed token
Wandb `wandb`	text (33)	`WANDB_API_KEY`	Key only
Baseten `baseten`	text (32)	`BASETEN_API_KEY`	Key only
Cloudflare `cloudflare`	text (30), transcripts (2)	`CLOUDFLARE_API_KEY`	Not yet needs your Cloudflare account id
Watsonx `watsonx`	text (30), transcripts (1)	`WATSONX_API_KEY`	Not yet needs your watsonx project and region address
Ollama (local) `ollama`	text (29)	`OLLAMA_API_KEY`	Not yet runs on your computer: needs its server address (no field for it yet)
Voyage `voyage`	embeddings (23)	`VOYAGE_API_KEY`	Key only
Anthropic (Claude) `anthropic`	text (22)	`ANTHROPIC_API_KEY`	Key only
Chatgpt `chatgpt`	text (22)	`CHATGPT_API_KEY`	Not yet signs in with a ChatGPT account, not an API key
Lambda Ai `lambda_ai`	text (20)	`LAMBDA_AI_API_KEY`	Key only
Volcengine `volcengine`	text (11), video (4), embeddings (5)	`VOLCENGINE_API_KEY`	Key only
Z.ai (GLM) `zai`	text (17)	`ZAI_API_KEY`	Key only
Hyperbolic `hyperbolic`	text (16)	`HYPERBOLIC_API_KEY`	Key only
Nscale `nscale`	text (14), pictures (2)	`NSCALE_API_KEY`	Key only
Tencent `tencent`	text (13), pictures (3)	`TENCENT_API_KEY`	Key only
Ovhcloud `ovhcloud`	text (15)	`OVHCLOUD_API_KEY`	Key only
Scaleway `scaleway`	text (12), transcripts (1), embeddings (2)	`SCALEWAY_API_KEY`	Key only
Runway `runwayml`	voice (1), pictures (2), video (11)	`RUNWAYML_API_KEY`	Key only
Aiml `aiml`	pictures (13)	`AIML_API_KEY`	Key only
Gradient Ai `gradient_ai`	text (13)	`GRADIENT_AI_API_KEY`	Key only
Ai21 `ai21`	text (12)	`AI21_API_KEY`	Key only
ElevenLabs `elevenlabs`	voice (9), transcripts (3)	`ELEVENLABS_API_KEY`	Key only
Libertai `libertai`	text (11), embeddings (1)	`LIBERTAI_API_KEY`	Key only
MiniMax `minimax`	text (6), voice (4), video (2)	`MINIMAX_API_KEY`	Key only
Moonshot (Kimi) `moonshot`	text (12)	`MOONSHOT_API_KEY`	Key only
Groq `groq`	text (7), voice (2), transcripts (2)	`GROQ_API_KEY`	Key only
DeepSeek `deepseek`	text (10)	`DEEPSEEK_API_KEY`	Key only
Morph `morph`	text (10)	`MORPH_API_KEY`	Key only
Tensormesh `tensormesh`	text (10)	`TENSORMESH_API_KEY`	Key only
Cerebras `cerebras`	text (9)	`CEREBRAS_API_KEY`	Key only
Meta `meta`	text (7), pictures (1), transcripts (1)	`META_API_KEY`	Key only
Publicai `publicai`	text (9)	`PUBLICAI_API_KEY`	Key only
Stability AI `stability`	pictures (9)	`STABILITY_API_KEY`	Key only
Cohere `cohere`	text (1), embeddings (7)	`COHERE_API_KEY`	Key only
SambaNova `sambanova`	text (8)	`SAMBANOVA_API_KEY`	Key only
Soniox `soniox`	transcripts (8)	`SONIOX_API_KEY`	Key only
Friendliai `friendliai`	text (7)	`FRIENDLIAI_API_KEY`	Key only
Gigachat `gigachat`	text (3), embeddings (4)	`GIGACHAT_API_KEY`	Key only
Cohere Chat `cohere_chat`	text (6)	`COHERE_CHAT_API_KEY`	Key only
Pinstripes `pinstripes`	text (6)	`PINSTRIPES_API_KEY`	Key only
Sagemaker `sagemaker`	text (6)	`SAGEMAKER_API_KEY`	Not yet needs an AWS access key pair, a region and your endpoint
Text-Completion-Openai `text-completion-openai`	text (6)	`TEXT_COMPLETION_OPENAI_API_KEY`	Key only
Lemonade `lemonade`	text (5)	`LEMONADE_API_KEY`	Not yet runs on your computer: needs its server address (no field for it yet)
Amazon Nova `amazon_nova`	text (4)	`AMAZON_NOVA_API_KEY`	Key only
Amazon Polly `aws_polly`	voice (4)	`AWS_POLLY_API_KEY`	Not yet needs an AWS access key pair and a region
Black Forest Labs (FLUX) `black_forest_labs`	pictures (4)	`BLACK_FOREST_LABS_API_KEY`	Key only
Heroku `heroku`	text (4)	`HEROKU_API_KEY`	Key only
Meta Llama `meta_llama`	text (4)	`META_LLAMA_API_KEY`	Key only
Azure Text `azure_text`	text (3)	`AZURE_TEXT_API_KEY`	Not yet needs your Azure endpoint and API version as well as the key
V0 `v0`	text (3)	`V0_API_KEY`	Key only
Assemblyai `assemblyai`	transcripts (2)	`ASSEMBLYAI_API_KEY`	Key only
Codestral `codestral`	text (2)	`CODESTRAL_API_KEY`	Key only
Darkbloom `darkbloom`	text (2)	`DARKBLOOM_API_KEY`	Key only
Featherless Ai `featherless_ai`	text (2)	`FEATHERLESS_AI_API_KEY`	Key only
Inception `inception`	text (2)	`INCEPTION_API_KEY`	Key only
Nlp Cloud `nlp_cloud`	text (2)	`NLP_CLOUD_API_KEY`	Key only
Recraft `recraft`	pictures (2)	`RECRAFT_API_KEY`	Key only
Text-Completion-Codestral `text-completion-codestral`	text (2)	`TEXT_COMPLETION_CODESTRAL_API_KEY`	Key only
Xiaomi Mimo `xiaomi_mimo`	text (2)	`XIAOMI_MIMO_API_KEY`	Key only
Text-Completion-Inception `text-completion-inception`	text (1)	`TEXT_COMPLETION_INCEPTION_API_KEY`	Key only

Some LiteLLM providers publish no model list (OpenRouter, for example). They are not offered in the model picker yet.

Previous← Connect your AI NextMCP tools →