Cohere Labs · Apache 2.0

Cohere Transcribe

Cohere’s 2B speech recognition for 14 languages, splitting long audio automatically.

How much memory it needs

Memory to run each model, from its model card or the size of its files. Smaller versions trade some quality for less memory.

ModelVersionNeedsBusiness use
Cohere Transcribe (03-2026)BF164.1 GBAllowed

Which computers run it

The best model in this group for each machine.

ComputerBest versionFit
Laptop or PC, no graphics card, 32 GBCohere Transcribe (03-2026) BF16Runs great
RTX 4060 · 8 GBCohere Transcribe (03-2026) BF16Runs great
RTX 3060 · 12 GBCohere Transcribe (03-2026) BF16Runs great
RTX 5060 Ti · 16 GBCohere Transcribe (03-2026) BF16Runs great
RTX 4090 · 24 GBCohere Transcribe (03-2026) BF16Runs great
RTX 5090 · 32 GBCohere Transcribe (03-2026) BF16Runs great
Mac with M4 · 16 GBCohere Transcribe (03-2026) BF16Runs great
Mac with M4 Pro · 48 GBCohere Transcribe (03-2026) BF16Runs great
Mac with M5 Max · 128 GBCohere Transcribe (03-2026) BF16Runs great
Ryzen AI Max+ 395 · 128 GBCohere Transcribe (03-2026) BF16Runs great
DGX Spark · 128 GBCohere Transcribe (03-2026) BF16Runs great
RTX PRO 6000 · 96 GBCohere Transcribe (03-2026) BF16Runs great
Mac Studio with M5 Ultra · 256 GBCohere Transcribe (03-2026) BF16Runs great
Mac Studio with M3 Ultra · 512 GBCohere Transcribe (03-2026) BF16Runs great

What it is good and bad at

Good at

  • Long recordings

Not good at

  • Download needs a (quick) approval

Start it

  • Cohere Transcribe (03-2026): runs in Transformers (CohereAsrForConditionalGeneration). Model card