Frequently Asked Questions

The Onysoft API is compatible with the OpenAI (Chat Completions and Responses) and Anthropic Messages API formats. The difference is that Onysoft offers models from many makers through a single API and a single key, and is paid in Turkish Lira.

Yes. In most cases you only need to change the base_url and api_key values and put a catalog id in provider/model form in the model field. For the forwarded request fields and known differences, see the Chat Completions section.

You can use models from OpenAI (GPT-6 Astra/Sol/Luna, GPT-5.6 Sol/Terra/Luna), Anthropic (Claude Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5), Google (Gemini 4 Argon, Gemini 3.8 Flash, Gemini 3.1 Pro preview), xAI (Grok 4.7, Grok 4.20), Meta (Muse Spark 1.3, Llama 4), DeepSeek (V4.1 Flash, V4 Pro), Qwen (Qwen3.8 Max Prime, Qwen3.8 Flash), Mistral (Mistral Large) and many more makers. The current list is on the Models page.

Yes. Claude Code works through the Anthropic Messages API endpoint (ANTHROPIC_BASE_URL=https://api.onysoft.com), Codex CLI through the Responses API endpoint (base_url = "https://api.onysoft.com/v1", wire_api = "responses") and Cursor with the "Override OpenAI Base URL" setting. Steps: Claude Code, Codex CLI, Cursor.

You can top up your balance in Turkish Lira by credit card or bank transfer from the Dashboard. A prepaid system is used.

Yes. For your top-ups an e-Invoice (corporate) or an e-Archive invoice (individual) is issued based on your billing details; the e-Invoice is delivered to your e-invoice inbox through the Revenue Administration (GİB) and the e-Archive invoice is sent to your email address.

You can get 402 (Payment Required) before your balance reaches zero: before every request the estimated maximum cost is compared with your balance (see Error Codes → Balance pre-check). You are charged for actual usage. Service resumes automatically after you top up.

Before every request the estimated maximum cost (input + max_tokens, or the model's maximum output if not sent, with a 20% safety buffer) is compared with your balance; if the balance is below it, 402 is returned. You are charged for actual usage. Lower max_tokens or top up; details in the Balance pre-check section.

/v1/chat/completions works synchronously and is for text/chat models (GPT-4, Claude, Gemini, Llama, etc.). The response returns immediately. /v1/video/generate, on the other hand, works asynchronously and is used for video, image and music generation — it first returns a task_id, then you poll the status with /v1/video/status/{task_id}.

Gemini image models (such as google/gemini-2.5-flash-image) are served by Google in the chat completions format — a base64 image is returned within the response. That is why they are called through /v1/chat/completions rather than /v1/video/generate.

i2i models (e.g. seedream/4.5-edit-i2i, ideogram/v3-edit-quality-i2i) require a source image. You can send a public HTTP URL or a base64 data URI with the image_url parameter. If you send base64, the system automatically saves it to a temp file and converts it to a public URL.

This error usually comes from the generation service and means required parameters are missing. The system fills in the required parameters for most models automatically, but some models (seedream, gpt-image, flux-2) require special parameters. The current code fills these parameters in automatically — if you still get this error, the model may be temporarily unavailable upstream. Try another model (e.g. google/imagen4-t2i or ideogram/v3-text-to-image-quality).

Image generation: 10-30 seconds. Video generation: 30 seconds - 3 minutes (depending on the model and the length). Music generation: 30-90 seconds. Poll the status endpoint every 5-10 seconds to track progress.

The GET /v1/models endpoint returns all active models with the prices for your key. Alternatively, you can visit the Models page in the Dashboard.

New models are added to the catalog through automatic synchronization. They typically become available within 24-48 hours of their public release.