Availability of Pay-As-You-Go Models in Unique

Overview

This page provides an overview of the pay-as-you-go models currently available within Unique. It highlights which models are accessible globally (all regions), specifically for Standard SWE, or for Standard CH.

The information is based on this Azure reference and augmented with our internal deployment knowledge.

Notes

Global Models: These models perform inference (data processing) across all regions provided by Microsoft, including the US. This is the default option for US tenants and can also be deployed in QA and OleOle for testing.
Standard SWE Models: Available exclusively in Sweden Central. This option applies when customers opt for data processing solely within Europe.
Data zone Standard Models: Data is processed somewhere in Europe (and not anymore only in Sweden). This is another option that applies when customers opt for data processing solely within Europe.
Standard CH Models: Designed specifically for Switzerland. This option applies when customers choose to have data processing only within Switzerland. This is the default option for all our Europe Single Tenants.
Request-Based Models: Certain models require a request and approval from Microsoft. (Refer to the Request Table below for details.)
PTU (Prepaid Through Unit): Some customers have purchased PTUs and maintain direct communication with Microsoft. However, they likely also follow a pay-as-you-go strategy, making this page relevant to them as well.
Escalations to Microsoft: If a model is not available, customers should escalate the request directly to Microsoft via the email contact they have from MS. The customer should submit a request specifying the model needed.

Model Availability Matrix

Model Name (can be copy pasted into the configs)	Available in code with release, still needs provisioning	Supported Base Models for UniqueAI Agent ❌: Not suited 🟡: functionally tested works ✅ Extensive Quality tests done	Available Globally, Only applicable for us-tenant Quota [tokens/min]	Data Zone Standard EU	Standard SWE Quota [tokens/min]	Standard CH Quota [tokens/min]	Needs Request for Standard Deployment? (Only ST and CMT)	URL to Request	Available on MT? Standard deployment only	Retirement (link)	Additional information

Model Name (can be copy pasted into the configs)	Available in code with release, still needs provisioning	Supported Base Models for UniqueAI Agent ❌: Not suited 🟡: functionally tested works ✅ Extensive Quality tests done	Available Globally, Only applicable for us-tenant Quota [tokens/min]	Data Zone Standard EU	Standard SWE Quota [tokens/min]	Standard CH Quota [tokens/min]	Needs Request for Standard Deployment? (Only ST and CMT)	URL to Request	Available on MT? Standard deployment only	Retirement (link)	Additional information
AZURE_GPT_35_TURBO_0125	before 2025.12	❌	❌	❌	❌	✅ 400K	❌	❌	✅	No earlier than July 16, 2025
AZURE_GPT_4_0613	before 2025.12	❌	❌	❌	❌	✅ 90K	❌	❌	✅	June 6, 2025
AZURE_GPT_4_32K_0613	before 2025.12	❌	❌	❌	❌	✅ 170K	❌	❌	✅	June 6, 2025
AZURE_GPT_4_TURBO_2024_0409	before 2025.12	❌	✅	❌	✅ 50k	❌	❌	❌	✅	June 6, 2025
AZURE_GPT_4o_2024_0513	before 2025.12	🟡 (use AZURE_GPT_4o_2024_1120 as replacement)	✅	❌	✅ 500k	❌	❌	❌	✅
AZURE_GPT_4o_2024_0806	before 2025.12	🟡 (use AZURE_GPT_4o_2024_1120 as replacement)	✅	❌	✅ 200k	❌	❌	❌	✅
AZURE_GPT_4o_2024_1120	2025.18	✅	✅	❌	❌	✅ 1M	❌	❌	✅
AZURE_GPT_4o_MINI_2024_0718	before 2025.12	❌	✅	❌	✅ 700k	❌	❌	❌	✅
AZURE_o1_2024_1217	2025.14	✅	✅ 2.5M	❌	❌	❌	✅	❌	✅		The system prompt is not allowed. The `temperature` and `topP` is always set to 1, —these values are hardcoded in the backend See: https://github.com/Unique-AG/monorepo/pull/11597/files .
AZURE_o1_MINI_2024_0912	before 2025.12	❌	✅	❌	✅ 60K	❌	✅		✅		The system prompt is not allowed. The `temperature` and `topP` is always set to 1, —these values are hardcoded in the backend See: https://github.com/Unique-AG/monorepo/pull/11597/files .
AZURE_o3_MINI_2025_0131	2025.14	❌ (no image processing)	✅ 2.5M	✅	❌	❌	✅		❌		The system prompt is not allowed. The `temperature` and `topP` is always set to 1, —these values are hardcoded in the backend See: https://github.com/Unique-AG/monorepo/pull/11597/files .
AZURE_GPT_45_preview	2025.12		✅	❌	❌	❌	✅	https://customervoice.microsoft.com/Pages/ResponsePage.aspx?id=v4j5cvGGr0GRqy180BHbR7en2Ais5pxKtso_Pz4b1_xUQTVJMVdCODFMUEhISlpGUVI2M1JEWEtTTSQlQCN0PWcu	❌
AZURE_GPT_41_2025_0414	2025.14	🟡	✅ 5M	✅ 2M	❌	❌	❌		❌
AZURE_GPT_41_MINI_2025_0414	2025.24		✅ 5M	❌	❌	❌	❌		❌
AZURE_GPT_41_NANO_2025_0414	2025.24		✅ 5M	✅ 2M	❌	❌	❌		❌
AZURE_o3_2025_0416	2025.20	✅	✅ 5M	❌	❌	❌	❌		❌
AZURE_o4_MINI_2025_0416	2025.20	🟡	✅ 5M	❌	❌	❌	❌		❌
litellm:openai-o1	2025.20	✅	✅	❌	❌	❌	❌		❌
litellm:openai-o3	2025.20	✅	✅	❌	❌	❌	❌		❌
litellm:openai-o4-mini	2025.20	🟡	✅	❌	❌	❌	❌		❌
litellm:openai-gpt-4-1-mini	2025.20		✅	❌	❌	❌	❌		❌
litellm:openai-gpt-4-1-nano	2025.20		✅	❌	❌	❌	❌		❌
litellm:openai-o3-pro	2025.26	❌: Chat Completion API is not supported for o3-pro. Only response API.	✅	❌	❌	❌	❌		❌
litellm:anthropic-claude-3-7-sonnet-thinking	2025.20	🟡	✅	❌	❌	❌	❌		❌
litellm:anthropic-claude-3-7-sonnet	2025.20	🟡	✅	❌	❌	❌	❌		❌
litellm:gemini-2-5-flash-preview-04-17	2025.20	🟡	✅	❌	❌	❌	❌		❌
litellm:llama-3-3-70b-instruct-turbo	2025.20	🟡 no forced tool calls possible web search creates a string instead of a tool call	✅	❌	❌	❌	❌		❌
litellm:deepseek-r1	2025.20	❌	✅	❌	❌	❌	❌		❌

Request Process for Restricted Models

If a model requires a request, follow these steps:

Submit a Request: Open a ticket in [Jira/ServiceNow] with details on why the model is needed.
Approval Workflow: The request will be reviewed by the Data Science Team, who will decide whether the client should make the request themselves or if Unique should do it
On Approval: A request is made on behalf of the client for our managed single tenants customer.
Access Provisioning: Upon approval, provision for the client using our terraform and tests on next, qa, prod and OleOle.

Author	@Pascal Hauri