The GPT-5 lineup: Pro, Mini, and Nano, and what to use in Yenisei at each tier
Three OpenAI models on one architecture: which GPT-5 to use for which workload, and which open models in Yenisei cover the same tiers.
Yenisei team
The GPT-5 lineup: Pro, Mini, and Nano, and what to use in Yenisei at each tier
GPT-5 Pro, GPT-5 Mini, and GPT-5 Nano are three OpenAI models on one architecture. Three tiers exist so you can match the model to your budget and latency needs.
There are no closed GPT models in the Yenisei catalog. From OpenAI we have the open GPT-OSS 120B and GPT-OSS 20B, and the same three tiers can be assembled from open models; below we show how.
What the GPT-5 lineup has in common
- Reasoning: in all three, differing in depth and consistency.
- Images as input: screenshots, photos of documents, diagrams.
- Tool calling: convenient for agents and integrations.
- 400K-token context: a solid middle ground between the classic 128K and a million.
How they differ
Pro is the top tier: architecture decisions, untangling convoluted logic, analytics with conclusions. Mini is the middle: for high-volume tasks that the lightest model can't handle. Nano is the fastest and cheapest: classification, short answers, field extraction.
Three tiers in Yenisei
- Top tier: DeepSeek V4 Pro (
deepseek-pro) or Kimi K3 (kimi-k3); our own alternative is Yenisei Max (yenisei-max). - Middle tier: GPT-OSS 120B (
gpt-oss-120b), OpenAI's open model with reasoning, or DeepSeek V4 Flash (deepseek-flash). - Light tier: GPT-OSS 20B (
gpt-oss-20b) or Mistral Nemo (mistral-nemo).
If you need images as input, at the middle tier take Qwen3.8 Flash (qwen-flash) or Gemma 4 31B (gemma-4-31b).
Switching is one line
curl https://api.yenisei.ru/v1/chat/completions \ -H "Authorization: Bearer $YENISEI_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "gpt-oss-120b", "messages": [{"role": "user", "content": "Summarize this email in three bullet points."}] }'Change gpt-oss-120b to deepseek-pro or gpt-oss-20b, and that's it. One OpenAI-compatible API, payment in rubles, 1 credit = 1 ₽.
How to choose in practice
Start with the middle tier. If the model consistently fails on your class of tasks, move up to the top tier. If the task is simple and there is a lot of it, move down to the light tier and save.