Callable Model Index
What a minute of AI calling really costs.
Every LLM, transcriber, voice and phone line a Callable agent can run on, priced per minute of conversation. These are the prices Callable charges, platform fee included, read from the same tables your balance is billed from.
Prices checked 10 November 2026
Your stack, per minute
$0.1757
$175.72 per 1,000 minutes
- LLM$0.0031
- Transcriber$0.0086
- Voice$0.1500
- Calling$0.0140
Price for one minute of conversation: the LLM reads about 10,000 tokens and writes about 100. Response time is the median on real Callable calls.
| # | Unit price | |||
|---|---|---|---|---|
| 1 | Gemini 2.5 Flash-Lite Google · Cheapest | $0.0021 | $0.20 in · $0.80 out per 1M tokens | Measuring |
| 2 | GPT-6 Luna OpenAI · Economy, lowest cost | $0.0021 | $0.20 in · $1.00 out per 1M tokens | Measuring |
| 3 | GLM-5.3 Flash Z.ai · Fast & cheap — best for voice | $0.0031 | $0.30 in · $1.00 out per 1M tokens | Measuring |
| 4 | GPT-4o mini OpenAI · Budget, reliable default | $0.0031 | $0.30 in · $1.20 out per 1M tokens | Measuring |
| 5 | Anthropic Haiku 5.5 Anthropic · Fast and affordable | $0.0032 | $0.30 in · $1.50 out per 1M tokens | Measuring |
| 6 | GLM-4.5 Air Z.ai · Budget | $0.0042 | $0.40 in · $2.20 out per 1M tokens | Measuring |
| 7 | GPT-5.4 Nano OpenAI · Fast replies | $0.0043 | $0.40 in · $2.50 out per 1M tokens | Measuring |
| 8 | Gemini 3.1 Flash-Lite Google · Fast and affordable | $0.0053 | $0.50 in · $3.00 out per 1M tokens | Measuring |
| 9 | Gemini 3.5 Flash-Lite Google · Balanced default | $0.0065 | $0.60 in · $5.00 out per 1M tokens | Measuring |
| 10 | GPT-5.4 Mini OpenAI · Advanced, smartest | $0.0159 | $1.50 in · $9.00 out per 1M tokens | Measuring |
| 11 | Anthropic Sonnet 5.5 Anthropic · Smartest, for nuanced conversations | $0.0630 | $6.00 in · $30.00 out per 1M tokens | Measuring |
| 12 | Anthropic Sonnet 5 Anthropic · Smart and balanced | $0.0630 | $6.00 in · $30.00 out per 1M tokens | Measuring |
Method
How the numbers are worked out
Every figure on this page comes from the same price tables your bill uses, measured against a real phone minute.
A minute of conversation
Each reply re-reads the agent's instructions and the conversation so far. Transcribers and phone lines are charged for every minute the call is connected. Longer instructions or a chattier agent cost more.
- agent replies
- ≈5
- agent replies a minute
- tokens the LLM reads
- 10,000
- tokens the LLM reads a minute
- tokens it writes
- 100
- tokens it writes a minute
- characters spoken
- 500
- characters spoken a minute
Callable prices
What comes out of your Callable balance for that part of the call, platform fee included. Computed from the same price tables the usage meter bills with, so the page and your bill can't disagree.
Response time
The median time from the caller's last word to the agent's first sound, on real Callable calls in the last 30 days. It includes transcription and voice, not just the LLM. A model shows a number after 25 calls; until then it says Measuring.
When prices change
Providers change their list prices. We check them regularly and the date at the top of this page says when we last did. A change applies to calls made after it, never to calls already made.
Start today
Build an agent on any of these.
Pick the stack that fits your calls. Switch models any time.
Live in under 10 minutes · Cancel any time