Embed 5 Fast

Not yet independently verified. Every detail here comes from Cohere’s own announcement and documentation; no independent report or measurement of Embed 5 was found. Its benchmark results are Cohere’s own. We will update this when it can be confirmed, and remove this note.

Embed 5 Fast is the smaller and cheaper of the two Embed 5 models Cohere released on 30 September 2026. According to Cohere’s announcement it shares Embed 5 Pro’s inputs (text, images, fused text and images, and parsed documents), its coverage of more than 100 languages, its 128,000-token context, its choice of 2048 to 256 output dimensions and its float, int8 and binary output types, and is available through the same channels: Cohere’s API and Model Vault, Microsoft Foundry, Amazon SageMaker, the North platform and private deployment. Cohere prices it at $0.08 per million text tokens and $0.40 per million image tokens. It is aimed at high-volume indexing where cost and latency matter more than the last points of retrieval quality. Cohere has not published separate evaluation results for it that we could check.