Model-provider realtime APIs - verified
OpenAI Realtime API (gpt-realtime, GPT-Live)
OpenAI Realtime API is listed under model-provider realtime APIs. OpenAI's Realtime guide tells anyone with a beta Realtime integration to migrate it to the GA interface before moving forward with new work. No date given. Its first listed price, in its own words: gpt-realtime-2.1 audio: $32.00 input, $0.40 cached input, $64.00 output per 1M tokens (per 1M audio tokens). Telephony stated on its pages: sip. Read from the vendor's own pages on October 20, 2026.
Last verified: 2026-10-20
Read from developers.openai.com. Docs - Pricing - OpenAI speech models - Compare every voice agent platform
What changed
- Beta interface: move to GAOpenAI Realtime API (gpt-realtime, GPT-Live)
OpenAI's Realtime guide tells anyone with a beta Realtime integration to migrate it to the GA interface before moving forward with new work. No date given.
“If you still have a beta Realtime integration, migrate it to the GA interface before moving forward with new work.”
developers.openai.com
- Status, as stated
GA interface; beta integrations told to migrate
“If you still have a beta Realtime integration, migrate it to the GA interface before moving forward with new work.”
developers.openai.com - Last verified
How OpenAI Realtime API charges
Every price line OpenAI Realtime API states, in its own words and units, with what it says the price includes and the quote. Nothing is converted.
| Price, in the vendor's words | Unit | Includes, as stated | Quote and source |
|---|---|---|---|
| gpt-realtime-2.1 audio: $32.00 input, $0.40 cached input, $64.00 output per 1M tokens | per 1M audio tokens | Not stated | “Prices per 1M tokens unless noted. · Model · Modality · Input · Cached input · Output / cost; gpt-realtime-2.1 · Audio · $32.00 · $0.40 · $64.00” |
| gpt-realtime-2.1-mini audio: $10.00 input, $0.30 cached, $20.00 output | per 1M audio tokens | Not stated | “gpt-realtime-2.1-mini · Audio · $10.00 · $0.30 · $20.00” |
| gpt-live-1: $0.05 per minute | per minute | See the GPT-Live billing note | “Model · Price per minute; gpt-live-1 · $0.05” |
| GPT-Live billing note | note | Billed per second; backend model and tool usage charged separately (stated) | “voice sessions are billed per second, without rounding up to a whole minute. Backend model and tool usage is charged separately.” |
- Free tier or credits
- Not stated
Not stated on the pricing page sections cited.
Telephony
- Phone numbers
- Not stated
No phone numbers offered on cited pages.
- SIP
Telephony and SIP guide listed in the Realtime docs navigationinferred
Inferred: Backed only by a navigation label in the Realtime docs ("Telephony and SIP"); the guide text does not itself describe SIP support.“Telephony and SIP”
developers.openai.com - Bring your own carrier
- Not stated
Not stated on the realtime guide.
Models, tools and calls
- Bring your own models
- Not stated
Single-vendor model API.
- Function calling and tools
Speech-to-speech model can call tools
“The model works directly with audio, maintains conversation state, and can call tools.”
developers.openai.com - Transfer and handoff
Handoffs inside the realtime session
“The agent handles audio turns, tools, interruptions, and handoffs inside that session.”
developers.openai.com - Recording and transcripts
- Not stated
Not stated on the realtime guide.
- Languages
- Not stated
Not stated on the realtime guide.
Self-hosting, license, compliance and scale
- Self-hosting
- Not stated
Hosted API only.
- License
- Not stated
Proprietary API.
- Compliance
- Not stated
Not stated on cited pages.
- Concurrency
- Not stated
Not stated on cited pages.
- SDKs
- Not stated
Guide uses the Agents SDK (@openai/agents/realtime) but does not list languages.
Vendor-stated latency
Latency as OpenAI Realtime API states it about itself. Vendor-stated, not measured.
- Vendor-stated
Low first-audio latency (no figure stated)
“This is the best starting point for voice agents that need barge-in, low first-audio latency, natural turn taking, and realtime tool use.”
developers.openai.com
Dataset note
Model-level facts for gpt-realtime-2.1, gpt-realtime-2.1-mini and gpt-live-1 are in the speech directory (provider key openai).
FAQ
What changed at OpenAI Realtime API?
OpenAI's Realtime guide tells anyone with a beta Realtime integration to migrate it to the GA interface before moving forward with new work. No date given.
How much does OpenAI Realtime API cost?
In OpenAI Realtime API's own terms: gpt-realtime-2.1 audio: $32.00 input, $0.40 cached input, $64.00 output per 1M tokens (per 1M audio tokens); gpt-realtime-2.1-mini audio: $10.00 input, $0.30 cached, $20.00 output (per 1M audio tokens); gpt-live-1: $0.05 per minute (per minute); GPT-Live billing note (note). As read on October 20, 2026; prices are not converted.
Does OpenAI Realtime API connect to phone calls?
Per its own pages: SIP: Telephony and SIP guide listed in the Realtime docs navigation.
Can I self-host OpenAI Realtime API?
Not stated. Hosted API only. License: not stated. Proprietary API.
Sources
Last verified from each vendor's own pricing pages, docs and license files; every quote was string-matched against the page that day. This is not a benchmark. Prices are shown in the vendor's own unit and change without notice, so check the linked source before you commit.
Newsletter
Know when a voice platform changes
One email per week. New model releases, agent tooling changes and what they mean for what you build.