Cohere

AI models: text generation, embeddings, reranking, and classification

Cohere API referenceAPI checked August 20, 2026

Signs in with: API token
Runs on: Cloud or On-Prem Worker
Operations: 26
Ready-made triggers: 0

Connect Cohere

In Neblex open Connect › Connections and click New Connection.
Search for Cohere and give the connection a clear name.
Fill in the fields. Secret values are encrypted and never shown again.
Click Test, then Create Connection.

Field	What to enter
Access tokenRequiredSecret

Use it in a flow

Open a flow, open the node library and search for Cohere.
Drag it onto the canvas and choose an Operation and your connection.
Map the operation's fields from earlier steps, or click Map with Matt.
Use Test this step to see what it returns, then save.

To start a flow from Cohere without a ready-made trigger: a Poll for changes trigger can check one of its list operations on a schedule and pass on only what is new; if Cohere can send webhooks, a Webhook trigger receives them at its own URL.

Operations

26 operations in 8 groups. Each one is a step you can add to a flow.

Audio1

Transcribe audiotranscribe POST Transcribe audio to text.

Batch4

Cancel a batch`cancelBatch`	POST	Cancel a batch job.
Create a batch`createBatch`	POST	Run a dataset of requests as a batch.
Get a batch`getBatch`	GET	A batch job's status.
List batches`listBatches`	GET	Your batch jobs.

Chat3

Chat (v1)`chatV1`	POST	Legacy v1 chat with connectors and RAG options.
Chat (v2)`chat`	POST	Generate a chat response from a messages array.
Preview prompt (v2)`promptPreview`	POST	The constructed prompt for a chat request, without generation.

Dataset5

Create a dataset`createDataset`	POST	Upload a dataset for embed jobs or batches.
Dataset usage`getDatasetUsage`	GET	Total dataset storage usage.
Delete a dataset`deleteDataset`	DELETE	Delete a dataset.
Get a dataset`getDataset`	GET	A dataset's details.
List datasets`listDatasets`	GET	Your datasets.

Embeddings6

Cancel embed job`cancelEmbedJob`	POST	Cancel a running embed job.
Create embed job`createEmbedJob`	POST	Embed a whole dataset asynchronously.
Embed (v1)`embedV1`	POST	Legacy v1 embeddings.
Embed (v2)`embed`	POST	Embed texts or images into vectors.
Get embed job`getEmbedJob`	GET	An embed job's status.
List embed jobs`listEmbedJobs`	GET	Your embed jobs.

Model2

Get a model`getModel`	GET	A model's details.
List models`listModels`	GET	Models available to the account.

Rerank2

Rerank (v1)`rerankV1`	POST	Legacy v1 rerank.
Rerank (v2)`rerank`	POST	Rank documents by relevance to a query.

Utility3

Check API key`checkApiKey`	POST	Validate the calling API key.
Detokenize`detokenize`	POST	Convert tokens back to text.
Tokenize`tokenize`	POST	Tokenize text with a model's tokenizer.