Violin is an inference engine, not another model. No parameter count, exact architecture or base checkpoint is published in this launch article.
Languages
Not collected
Launch benchmarks evaluate English and selected European languages. The current API’s ten-language contract and the benchmark row’s nine-language count are not assumed to be immutable release-specific coverage.
Context caching
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
Current v1 API does not expose a documented Resonant-1 or Resonant-1 Flash base-model selector. Named release access and any immutable snapshot pinning need provider confirmation.
Access
Official launch and benchmark reference. Contact Reson8 for exact named-release access; public v1 documentation automatically routes to the current pipeline and provides no immutable base-model selector.
Regions
Original benchmark deployment region not established
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Current terms are linked for context. The original release-specific processing agreement and current pinned-model access are unverified.
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.
Reson8 · named release, current routing unverified
Not verified
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
How it works
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
Minimum reusable input
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
Lifetime & refresh
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
How to enable it
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
Usage fields
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
Billing rules
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
Limits & data boundaries
Not collected
No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
What the measurements say
Launch English short-form · five-dataset mean
Reson8 · provider-reported
6.71% WER
April 8, 2026 provider table: AMI, Earnings22, Gigaspeech, SPGISpeech and Tedlium only. LibriSpeech and VoxPopuli excluded for training overlap. Not the later leaderboard aggregate or a universal accuracy score.
Provider-reported RTFx from launch comparison. Hardware, batching and end-to-end upload overhead are not fully specified. Not a production latency SLA.
Hugging Face Open ASR Leaderboard · independent benchmark
8.21% WER
Pinned benchmark-owner commit d3c0e9a07b475a64afb56254e8c35b584905af32, exact named model row and dataset column. Lower is better. Different cleaned/chunked variants are not interchangeable with launch datasets.
Hugging Face Open ASR Leaderboard · independent benchmark
5.48% WER
Pinned benchmark-owner commit d3c0e9a07b475a64afb56254e8c35b584905af32, exact named model row and dataset column. Lower is better. Different cleaned/chunked variants are not interchangeable with launch datasets.
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
The provider names and describes training these two releases, and the pinned benchmark-owner catalog explicitly marks each Reson8 release Proprietary. No public model weights were identified; SDK and inference configuration are not weight distributions.
