Embed 5 Fast

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.

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.