Ultravox

Ultravox is a real time voice AI platform for building speech native voice agents. Rather than converting speech to text before a language model reads it, Ultravox trains its own speech model that takes audio directly, which the vendor says preserves tone, cadence and pitch and removes the transcription step that adds latency. The company runs its own inference stack and infrastructure instead of relying on external model providers. The site states that Ultravox has been acquired by Inworld to accelerate its speech to speech research and voice agent capabilities. The audience is developers and teams building phone and conversational agents. The platform offers REST APIs, SDKs for major platforms, an unlimited playground, SIP calling, an outbound call scheduler, custom voice clones and retrieval augmented generation over uploaded corpora. A related Threads feature is billed by token. Note that the platform is a voice agent product and not a video editor, despite its catalog placement. Pricing is published. The Pay as You Go plan costs $0 per month and includes 30 free minutes of calls, then $0.05 per minute, with no surge pricing and up to five concurrent calls. Pro is $100 per month, shown as an annual rate, with the same per minute price, no hard concurrency cap, the outbound scheduler, five custom voices and 20 corpora. SIP calls are priced at 0.5 cents per minute on Pay as You Go. Threads usage is $2.00 per million uncached input tokens and $15.00 per million output tokens. Enterprise pricing is custom and adds a priority SLA and organization support.