September's new models in the catalog: Qwen3.8 2.4T, GLM 5.3, Inkling and more

Update: on September 3 the catalog grew by sixteen models. Some of them have since been withdrawn: the Yenisei catalog now holds open-weight models and our own Yenisei lineup. Below are the models from that release that are available now.

All of them work through the OpenAI-compatible API, with payment in rubles and only for the tokens you use. Prices are in rubles per 1M tokens as of the publication date; current prices are always on the model's page.

Large models

• Qwen3.8 2.4T, the largest model in the Qwen lineup: reasoning and a context of about a million tokens. 160 / 480 ₽.
• Inkling, the first model from Thinking Machines: text, images, audio, 1M context. 80 / 324 ₽.
• GLM 5.3, one of the largest contexts in the catalog, over 1.3M tokens. 112 / 352 ₽.

Fast and inexpensive

• GLM 5.3 Flash: 6 / 20 ₽ with a context over 1.3M and image and video input. The best combination of capacity and price in the catalog.
• Ling 3.0 Flash: 1.68 / 5.04 ₽, the most affordable of the new arrivals.
• DeepSeek V4 Flash Vision, a fast DeepSeek that accepts images as input. 17.6 / 52.8 ₽.

For code

• Qwen3 Coder Next: 9.6 / 64 ₽, 262K context. An inexpensive assistant for everyday work.

How to choose

Start with the cheap ones: if you need a short answer and have many requests, take Ling 3.0 Flash, it costs next to nothing. If you need a large volume of material in a single request, go with GLM 5.3 or GLM 5.3 Flash. If you need images or video, GLM 5.3 Flash or DeepSeek V4 Flash Vision. And only when the task is genuinely complex, take a large model: the price difference between Ling 3.0 Flash and Qwen3.8 2.4T is almost a hundredfold.

All models work through one key and one address, https://api.yenisei.ru/v1. The full list is in the catalog.