Release Notes

Model Vault and Model Vault Encrypted now run leading third-party open models alongside Cohere’s models, on fully managed, dedicated infrastructure. You can choose the right model for each task and run every one of them with the same protections.

Key features

• Open models in Model Vault: deploy models from DeepSeek, Z.ai (GLM), Xiaomi (MiMo), NVIDIA (Nemotron), Google (Gemma), OpenAI (GPT-OSS), Microsoft (Phi), Alibaba (Qwen), and Mistral AI. See Standard Vault supported models.
• Open models in Model Vault Encrypted: run Gemma, GPT-OSS, Phi, Qwen, Devstral, and Ministral with confidential computing, so not even Cohere or the cloud provider can see your prompts or responses. See Encrypted Vault supported models.
• Zero Data Retention by default: prompts and responses are not retained in Model Vault.
• Regional serving: Model Vault runs in the United States, the European Union, and Canada. Model Vault Encrypted runs in the United States, with the European Union and Canada coming soon. See Regions.

Calling open models

Open models are served through the OpenAI-compatible Chat Completions API at your vault endpoint:

from openai import OpenAI
client = OpenAI(
    api_key="<COHERE_API_KEY>",
    base_url="<YOUR_VAULT_ENDPOINT_URL>/v1",
)
response = client.chat.completions.create(
    model="deepseek-v4-1-flash",
    messages=[{"role": "user", "content": "Hello from Model Vault!"}],
)
print(response.choices[0].message.content)

See Calling a Standard Vault over the API and Calling an Encrypted Vault over the API.

Availability

Open models are available today in the Model Vault app. Model Vault is priced per GPU, per hour, for both Standard and Encrypted Vaults; see the pricing page. For reserved capacity, contact our sales team.