Integration guides ยท 2026-09-20
Using the EVREN API through LLMTR: your own key, one endpoint
EVREN models sit in the LLMTR catalog as five rows and run on your own EVREN key. How to register the key, what travels in the request, which error means what, and which features these rows do not offer.
Short answer
You can call EVREN models through LLMTR, and the key stays yours. Register the API key you created in the EVREN console with LLMTR once, then call the model from the same endpoint as every other model.
You can connect to EVREN directly; what connecting through LLMTR changes is that the same code also reaches other Turkey-hosted models and global models. Nothing changes except the model id.
The five EVREN rows in the catalog
EVREN publishes eleven models; the five its own table describes as chat models are in the LLMTR catalog. The embedding, reranking, OCR and speech recognition rows are not here yet.
Context and modality figures come from EVREN's own published model table.
| Model id | Context | Input |
|---|---|---|
| evren/deepseek-v4-flash-tr | 1,000,000 | Text |
| evren/glm-5.3-fp8 | 512,000 | Text |
| evren/gemma-4-31b | 256,000 | Text, image |
| evren/qwen3.8-flash-next | 256,000 | Text, image, video |
| evren/qwen3-vl-30b | 256,000 | Text, image, video |
Register your key once
Unlike every other model in the catalog, these rows do not work until you register an EVREN key. Registration happens once and is not repeated per request.
Once saved, the key is never shown again; only its last four characters remain visible. If you lose it, create a new one in the EVREN console and replace it in Settings.
- Create an API key in the EVREN console and accept EVREN's terms of use.
- Paste the key into the provider keys card in LLMTR Settings.
- Call the model as usual; only your LLMTR key travels in the request.
First request to an EVREN row
curl "$LLMTR_BASE_URL/v1/chat/completions" \
-H "Authorization: Bearer llmtr-your_key" \
-H "Content-Type: application/json" \
-d '{
"model": "evren/glm-5.3-fp8",
"messages": [
{ "role": "user", "content": "Bu sozlesmedeki fesih maddelerini ozetle." }
],
"max_tokens": 4096
}'What travels in the request, and what does not
Your request's authorization header carries only your LLMTR key. You do not put your EVREN key in the body or in a header; it stays where you registered it and is used only to send your own request.
The key is stored with strong encryption and stops working the moment you remove it. It is never shown to you again either.
Request and response bodies are not written to our database. The usage record holds the model, token counts, message count, cost, status code and duration. The third-party model provider that processes the request may keep records, and every model card says so; on EVREN rows that provider is EVREN itself, and the account is yours.
Error responses
On these rows the provider's answer reaches you directly. Because both the key and the account are yours, you are the only person who can fix a problem.
The details.upstreamStatus field always carries the original code EVREN returned. This distinction matters because your clients decide on retries from the HTTP status: 4xx codes are preserved as they are, and 5xx responses are collapsed into 502.
| Situation | HTTP | error.type |
|---|---|---|
| You have no registered key | 403 | provider_credential_required |
| EVREN rejected your key | 403 | provider_credential_rejected |
| Your EVREN account did not cover the request | 403 | provider_account_limited |
| EVREN applied a rate limit | 429 | provider_rate_limited |
| EVREN refused for another reason (4xx) | the provider's code | provider_upstream_error |
| A fault on EVREN's side (5xx) | 502 | provider_upstream_error |
What these rows do not offer
Tool calling, JSON mode and thinking controls are not offered on these rows. Do not send those parameters.
Sending a tool definition to a row that does not offer it may not return an error: the model can write in prose what the tool should have returned, and the request answers 200. For a step that needs tool calling, pick a catalog row that carries the capability.
Streaming on these rows is not real streaming: the response is delivered in pieces after it completes. No separate output ceiling is published, so the limit on the model card is the context window.
Why come through here instead of going direct?
Going direct works too, and this article does not claim otherwise. The difference shows up on the day you reach for a second model.
You can connect to EVREN directly; what connecting through LLMTR changes is that the same code also reaches other Turkey-hosted models and global models. The catalog holds twelve live Turkey-hosted rows today, seven of them outside EVREN, and moving to a global model changes nothing but the model id in the request.
LLMTR does not charge for these requests. The 8 percent platform share added when you load credit is unchanged; it applies to credit loading, not to model rates.
Calling EVREN models through LLMTR
Steps for registering your EVREN API key with LLMTR and calling EVREN models from the OpenAI-compatible Chat Completions endpoint.
- Create an EVREN key. Create an API key in the EVREN console and accept EVREN's terms of use.
- Register the key with LLMTR. Paste the key into the provider keys card in LLMTR Settings and save it.
- Call the model. Send a request to the Chat Completions endpoint with a model id beginning with evren; the authorization header carries only your LLMTR key.
- Read the error state. If a request fails, read error.type; the details.upstreamStatus field carries the original code EVREN returned.
Frequently asked questions
Should I use EVREN directly or through LLMTR?
You can connect to EVREN directly; what connecting through LLMTR changes is that the same code also reaches other Turkey-hosted models and global models. For a single model both routes give the same result; the difference opens up at the second model.
Does my EVREN key travel in my request?
No. Your request carries only your LLMTR key. Your EVREN key stays registered in Settings, is stored with strong encryption, and is used only to send your own request.
Does LLMTR charge for these requests?
No. The catalog price of EVREN rows is zero and these requests do not draw down your LLMTR balance. The 8 percent platform share on credit loading is separate and unchanged.
Does tool calling work on EVREN rows?
No, tool calling is not offered on these rows. Do not add tool definitions to your request: the answer can come back as prose rather than an error. For a step that needs tool calling, pick a catalog row that carries the capability.
What happens if I remove my registered key?
The key stops working the moment it is removed, and requests to that provider's rows stop with 403 provider_credential_required. Every other model in the catalog keeps working.