Provider model labels, backend selections and voice IDs do not establish an immutable request-side checkpoint selector.
Access
wss://relay.addisassistant.com/ws?apiKey=YOUR_API_KEY is the documented relay contract. Separate from one-use-ticket TTS and Scribe sessions. No live connection performed.
Regions
Policy allows Ethiopia, Switzerland, EEA and provider regions. Exact request routing and private residency require agreement.
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 16 Oct 2026
Category
Usage
Rate
Unit
Realtime audio
Realtime audio input1
ETB 1,000.00
1M audio tokens
Realtime audio
Realtime audio output2
ETB 4,000.00
1M audio tokens
Rate conditions
Native ETB: 1 ETB per 1,000 audio input tokens, normalized to 1M. Not ordinary text tokens or generated audio minutes.
Native ETB: 4 ETB per 1,000 audio output tokens, normalized to 1M. Do not feed audio pricing into the text-question estimator.
Standard-service terms/privacy permit content review, model training, fine-tuning and improvement. No-training or restricted processing requires a controlling written enterprise agreement. Account content and clips may persist for feature use, safety/legal needs or agreed terms; transient realtime processing has separately retained operational logs. No universal numeric retention TTL. Processing may occur in Ethiopia, Switzerland, EEA and provider regions; private residency requires written agreement.
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
Addis AI
Not verified
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
How it works
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
Minimum reusable input
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
Lifetime & refresh
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
How to enable it
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
Usage fields
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
Billing rules
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
Limits & data boundaries
Not collected
No native inference-cache contract reviewed. Request replay, durable clips, fileUri reuse and recovery windows are distinct from prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 Oct 2026
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Commercial hosted API with restricted service access and reserved provider/licensor IP. Public research artifacts exist; this entry is not an assertion of an independently closed-weight checkpoint.
