Is API.ai Free? What It Actually Costs Today

Short answer

API.ai isn't a product anymore. Google acquired it in 2016 and folded it into Dialogflow in late 2017. If you're looking for "api.ai" today, you'll land on Google's Dialogflow console instead — the old branding, dashboard, and API endpoints are gone.

So the real question is: is Dialogflow free? The answer is: partly. Dialogflow ES (the original, simpler agent-building product) has a genuinely usable free tier for text-based interactions. Dialogflow CX (the newer, more powerful version aimed at complex conversational flows) is pay-as-you-go from the first request, with no meaningful free allowance. Which one you're pointed at depends on when you created your project and which console you're using.

What API.ai used to be

Before the acquisition, API.ai was a standalone natural language understanding (NLU) tool. Developers used it to define intents, entities, and training phrases, then hook a chatbot or voice assistant up to it via API calls. It had a free tier and paid plans for higher volume and premium features like sentiment analysis add-ons.

Google absorbed the technology, renamed it Dialogflow, and rebuilt the pricing model around Google Cloud's billing system. If you still have old API.ai documentation bookmarked, treat it as historical — the endpoints, SDKs, and pricing structure described there no longer apply.

What's actually free in Dialogflow today

Dialogflow ES (Essentials):

• Text-based requests have a free monthly allowance before you're billed per request.
• Voice/audio requests (speech-to-text, text-to-speech through the API) are billed separately and don't share the same generous free tier.
• Standard agent building, intent matching, and basic integrations (web demo, some messaging platforms) are free to set up.

Dialogflow CX:

• No free tier for API requests — you're billed per session/request from the start, though Google occasionally offers Cloud-wide free credits for new accounts.
• CX is built for larger, more complex conversational applications, so the cost model assumes production usage rather than experimentation.

In practice: if you're prototyping a simple FAQ bot or basic voice assistant with text input, Dialogflow ES's free tier can carry you a long way before you see a bill. If you need CX's visual flow builder, state management, or you're doing heavy voice traffic, budget for it from day one.

Why people still search "is api.ai free"

Most developers asking this aren't trying to relive 2016 — they're looking for a free or cheap way to build conversational AI and found old tutorials, Stack Overflow answers, or YouTube videos referencing API.ai that never got updated. It's worth knowing the landscape has shifted in two ways since then:

1. Google consolidated the product into Dialogflow, with a more complex two-tier pricing structure (ES vs CX).
2. General-purpose LLM APIs (Claude, GPT-4, Gemini) have largely replaced narrow intent-based NLU for a lot of chatbot use cases. Instead of manually defining intents and training phrases, you can hand a well-crafted system prompt to a large language model and get more flexible, natural conversation handling with far less setup.

If your goal is "build something that understands user text and responds intelligently," a modern LLM API is often simpler to get started with than rebuilding an intent-matching system from scratch.

Free tiers vs practical free access

Almost every AI API — Dialogflow included — offers some free tier, but the free tier is rarely the constraint that matters in practice. What actually determines your cost is:

• How much traffic you send per month
• Whether you need voice, not just text
• Whether you're building one bot or serving it to a team/product with multiple developers

If you already pay for a Claude subscription for your own use, and you want to expose that same model through an API for a bot, script, or internal tool, you don't need to sign up for a separate NLU platform's paid tier. SubToAPI turns Claude access you already have into an HTTPS API with an application key, so you can call it from your own bot logic without adopting Dialogflow's intent-based model at all.

curl https://api.subtoapi.app/v1/messages \
  -H "Authorization: Bearer $SUBTOAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4",
    "max_tokens": 500,
    "messages": [
      {"role": "user", "content": "Book a table for two at 7pm"}
    ]
  }'

That's a full conversational request-response, no intent schema to design, no training phrases to maintain. See the quickstart and the messages docs for the full request/response shape, and streaming if your bot needs to display responses as they generate.

Dialogflow vs a raw LLM API: which to pick

• Pick Dialogflow (or a similar NLU platform) if you need tight, predictable intent classification for a small, well-defined set of actions — think "book appointment," "check order status," "cancel subscription" — and you want built-in integrations with phone/voice channels.
• Pick a general LLM API if your bot needs to handle open-ended questions, generate varied natural language, reason through multi-step requests, or use tools/function calling to take actions mid-conversation. SubToAPI's tool use support covers that last case if you're building something that needs to call external functions based on the conversation.

Neither is strictly "better" — they solve different problems. But if you started searching "is api.ai free" because you assumed Dialogflow was your only path to a conversational bot, it's worth knowing it isn't.

questions

Is API.ai still available as a separate product? No. It was renamed to Dialogflow after Google's 2016 acquisition. All new development, documentation, and billing happens under the Dialogflow name.

Does Dialogflow have a free tier? Dialogflow ES has a free monthly allowance for text requests; voice requests and Dialogflow CX are billed from the start, with occasional Google Cloud credits for new accounts.

What's a simpler alternative to building intent-based bots? Using a general-purpose LLM API directly — sending user text and getting a natural response — skips intent/training-phrase setup entirely. You can sign up and start with a free trial to see if that approach fits your bot.