GLM models
Mind the letter case in the IDs: the previous-generation model is uppercase (
GLM-5.2), while the 5.3 generation is lowercase (glm-5.3, glm-5.3-flash). Copy IDs from the pricing page as they are.glm-5.3 or glm-5.3-flash, for images use deepseek-flash, Claude or Gemini. Details are in the changelog.
Strong across languages
A key trait of GLM is confident multilingual coverage. The models handle Russian, English and Chinese equally well, which makes the family handy for mixed-language content: a Russian-language chatbot, translation, or parsing documents in several languages. The previous flagship GLM-5.2 gives the most even quality on hard multilingual tasks, while glm-5.3 and glm-5.3-flash are plenty for everyday Russian text without overpaying. For long answers you can stream the response token by token instead of waiting for the full reply. For agents and structured tool calls, see function calling.Working with images
Good value
GLM’s main draw is price. It’s one of the best-value families: capable models at a noticeably lower per-token cost, and for a high volume of simple requests there’s glm-5.3-flash, the cheapest in the family. We don’t pin numbers on this page — check the exact per-token price for each ID on the Pricing page at www.ruapi.ai. Top up in USDT — see the Top-up page.Connecting GLM
GLM is served over the OpenAI-compatible protocol — no separate library needed. Take any OpenAI SDK, change thebase_url to https://www.ruapi.ai/v1, and put the ID you want in the model field.
Next
Connect and first request
base_url, key, Python and curl code.
DeepSeek models
Same key — another strong value line.
Gemini models
Multimodal input and long context.