EVREN AI platform launched: what enterprises should know

EVREN AI platform launched: domestic GPUs, 11 models and a single API

On 27 September 2026 Türkiye’s Presidency of Defence Industries (SSB) announced that the national EVREN AI platform is open for use. According to the statement reported by Anadolu Agency, the platform brings data, AI models and high-performance computing together under one roof for defence companies, technology firms, academics, entrepreneurs and students. Users sign in at evren.ssyz.org.tr with e-Devlet, Türkiye’s national e-government login.

Open to academia since June, the platform both widened its audience and formally introduced its language model layer with this announcement. This news piece summarises the main points of the statement, what has changed since June and what it means for enterprise software teams. Aksiyon Soft has no partnership with SSB or EVREN; this is independent industry news.

In short

What: on 27 September 2026 SSB announced that EVREN is open to a broad audience from industry and academia.
Data: prompts, responses and usage data are processed on GPU infrastructure in Türkiye.
Capabilities: object detection, segmentation and classification; 11 open-weight language models; an API with real-time responses and automatic model routing.
Credits: credits earned by uploading datasets, labelling and sharing models are spent on training and language model inference.
Campaign: API calls are not deducted from credit balances until 1 November 2026; the platform has 7,500 active users.

What does the SSB statement say?

According to the statement, EVREN offers the entire AI development process on one platform. On the computer vision side, object detection, segmentation and classification models can be built with end-to-end labelling and training. On the language side, an inference layer hosts 11 open-weight large language models, and the API architecture supports real-time response streaming and automatic model routing.

The statement stresses data sovereignty: users’ prompts, responses and usage data are processed on GPU infrastructure in Türkiye. The credit model has also grown; users earn credits by uploading datasets, labelling data or sharing the models they have trained, and can spend those credits on model training and language model inference. The platform is meant to support the goals of Türkiye’s AI Action Plan, and future plans include domestic language, vision and audio models as well as distributed GPU management.

Diagram: where EVREN fits in an enterprise architecture — business apps, AI-ops control layer (KVKK masking, audit log), EVREN API, domestic GPUs — EVREN in an enterprise architecture: our suggested control layer for masking, permissions, audit logging and fallback between business apps and the EVREN API.

What has changed since June?

Dünya newspaper’s 28 September report recalls that e-Devlet access to the platform began in June and that the latest step added the language model service. The table below compares YÖK’s June letter with SSB’s September statement, using the information in those sources.

Topic	June 2026 (YÖK letter)	September 2026 (SSB statement)
Audience	academics and students	defence companies, technology firms, academics, entrepreneurs and students
Capabilities	MLOps lifecycle from data preparation to publishing	plus 11 open-weight language models, real-time responses and automatic model routing
Earning credits	uploading and labelling datasets	plus sharing trained models
Spending credits	model training on GPUs	model training and language model inference
Campaign	starter credit via QR code (until 31 July 2026)	API calls not deducted from credits until 1 November 2026

There is movement on the developer side too: the platform’s official Python package, evren-sdk, gained a language model client with chat, streaming, embedding and rerank calls in its 0.9 releases published on PyPI on 24 September 2026.

What does it mean for enterprise software teams?

In our assessment, the most important consequence for organisations is that a domestic AI infrastructure is no longer just an academic testing ground. Explicitly adding technology companies and entrepreneurs to the audience makes EVREN a concrete option for projects whose data must stay in Türkiye. Thanks to the OpenAI-compatible interface, trying it with existing applications is cheap.

On the other hand, domestic infrastructure does not remove responsibilities on the application side. Which data is sent to a model, who may use which model, how spending is tracked and what happens when the service does not respond must still be solved in the organisation’s own layer. Automatic model routing makes model selection easier inside the platform, but the organisation’s own rules (for example, which model receives prompts that contain personal data) should be defined in a control layer between the application and the platform.

Team working at a whiteboard — planning architecture and KVKK decisions for an EVREN AI platform integration — Agreeing on data classification, permissions and budget as a team before integration sets the pace of the pilot (illustrative image).

Enterprise evaluation checklist

Have you listed which use cases must keep their data in Türkiye?
Are masking rules written for personal data in prompts?
Are permissions and quotas defined per user, department and model?
Is an audit log (who, when, which model) kept for every call?
Is there a budget and alert threshold for credit consumption after 1 November 2026?
Is a backup model or provider ready for service outages?

EVREN series

This is the fourth instalment of our series following the EVREN platform. The series began with the release of EVREN SDK on PyPI, continued in June when the platform opened to universities through YÖK, and entered a new phase in September with the LLM inference service with 11 models. We explain how to connect the API to enterprise software in our EVREN API integration guide.

How can Aksiyon Soft help?

Our Samsun-based team works remotely with organisations across Türkiye and makes planned on-site visits when needed. Through our API and integration service we build the control layer between your business applications and AI services: masking, role-based permissions, audit logging, quotas and fallback. To connect several systems to the same services, our API and data integration platform solution provides a shared foundation.

The process starts with a discovery session, followed by a single-scenario MVP, two-week sprint demos and hypercare plus SLA-backed support after go-live. You can find our architectural approach in detail in our article on the AI-ops layer and LLM router.

Frequently asked questions

Who can use the EVREN AI platform?

According to the SSB statement, the platform is designed for defence companies, technology firms, academics, entrepreneurs and students. Sign-in is with e-Devlet at evren.ssyz.org.tr.

Are API calls paid?

Usage is credit-based. According to the statement, API calls are not deducted from credit balances until 1 November 2026. Credits are earned by uploading datasets, labelling and sharing models.

Where is data processed?

According to the statement, prompts, responses and usage data are processed on GPU infrastructure in Türkiye.

What does automatic model routing do?

The statement says the API architecture supports automatic model routing, meaning the platform directs a request to a suitable model. We still recommend your own control layer for organisation-specific rules.

How many people use the platform?

According to the statement reported by Anadolu Agency, the platform has 7,500 active users.

Is Aksiyon Soft an EVREN partner?

No. Aksiyon Soft has no partnership with SSB or EVREN. This article is industry news based on public sources.

Let’s talk about your project

If you want to connect domestic AI services to your business applications safely, write to us through our contact form and we will review your scenario and data classification together in the first call.

National AI platform EVREN launched in Türkiye