Whisper large-v3-turbo
safetensors · openai/whisper-large-v3-turbo
queued
| Publisher | openai |
|---|---|
| Modality | speech-to-text |
| Parameters | 0.809B 809M params (HF safetensors count). Source: model card. |
| Format | safetensors |
| Size on disk | 1.51 GiB (1,622,466,054 bytes) |
| Estimated memory to run | ~5.6 GB (estimate; see method) On the NAS this is CPU-only and slow (Intel N100). |
| Native context | n/a |
| Engine | Transformers / openai-whisper / whisper.cpp (own ggml file) · recipe |
| License | mit MIT: commercial OK. |
| Archive folder | media |
| Notes | 809M params (HF safetensors count). |
Where it runs
| Machine | Verdict | Budget used |
|---|---|---|
| Bosgame M5 (x1) | Fits | ~108-110 GB GPU-usable on Linux with GTT raised (96 GB cap on Windows). Fit budget used here: 108 GB. |
| Bosgame M5 (x2, RPC) | Fits | ~200-225 GB usable across two nodes. Fit budget used here: 210 GB; 210-225 GB = tight. |
| MacBook Pro (M4 Max) | Fits | macOS caps GPU wired memory by default (~75% = ~27 GB). Raising iogpu.wired_limit_mb to ~32 GB is possible but leaves little for the OS. |
| Mac mini (M4) | Fits | ~2/3 of 16 GB usable by the GPU by default. Already busy as build box + Hermes/Ollama. |
| omarchy (AMD desktop) | Fits | 7 GB of the 8 GB VRAM for weights. Offload capacity depends on system RAM, which is unknown today; after the 128 GB upgrade ~110 GB of MoE experts could sit in RAM. |
| Ugreen DXP2800 NAS | Tight / offload | Storage host. Already runs ollama/open-webui containers; only tiny models (e.g. Whisper on CPU) are realistic. |
| MacBook Neo | Unknown | Specs unknown: machine was offline/unreachable on Nov 14 2026. |
Token/s = theoretical ceiling from memory bandwidth and active weight bytes; not a benchmark. Nothing on this page is a benchmark score.