Whisper large-v3-turbo

Whisper large-v3-turbo · FP16 · by openai

openai/whisper-large-v3-turbo

openai · FP16 · on speech-hf · vendor build

Measured by 6 nodes on hardware we do not own, and signed by each of them. What machines here actually proved — never what the card claims.

matrix built 12 Nov 2026 · every field verified against the signature of the node that produced it, and re-verifiable by you: each signed figure carries the exact bytes the node signed (signing_payload_b64) and the signature over them, and the ed25519 public key is embedded in node_did. effective_ctx is the context length a node PROVED by recall, never the length a model card advertises

Run it on your gram Install a gram with one command. It measures models like this one on your own machine and signs every result.Use it in your own room Ask Evo in your gramx and it runs on a machine that proved it. On your own machines it costs you nothing.

context declarednot declaredwhat the runtime reported. A node attests it was told this; nobody attests it is true.

context proved by recallnot probedthe largest size at which a machine could still find tokens planted across the whole prompt — start, middle and end, all of which had to come back.

Both numbers are needed to state a gap, and one of them is missing.

Languages, proved

Bengali English Gujarati Hindi Kannada Marathi Nepali Tamil Telugu Urdu

Each question was put entirely in its own language and script, and the answer had to be right. A model with no coverage cannot parse the question, so answering in English is a failure here, not a pass. 10 proved of 13 asked — a language absent from the probe set was never asked, and is not a claim either way.

what it produced, measured

speech recognition

produced21stook37sspeed1.8× slower than realtime

measured on Apple Silicon, 8 cores, 16 GB · realtime_factor 1.78 (wall ÷ media)

The fastest measurement any machine reported, signed by the machine that took it. Every other machine’s numbers are in the table below — a rate is a fact about a model and a computer, and two machines disagreeing is two hardware classes rather than one of them being wrong.

Capabilities, measured

Speech inpassed

Whether the model itself accepts audio in a turn — distinct from this gram running a separate transcription engine. Recorded as unmeasured until an engine exposes the path, never as a failure.

signed by this network

What machines here proved

best throughput1.8× slower than realtime

resident while loaded3.7 GiB · did:epn:002408…e61d3b

machines that measured it6

regions it was proved in4

Every line traces to a node signature.

from the public record · unsigned

What the runtime says about it

No runtime metadata was attested for this model.The only unsigned block on this page. We attest we read these from the runtime — not that they are true. Nobody signs a model card.