Two new models on the API: GLM-5.3-Flash and MiMo-V2.6-Flash 🚀

Two new models on the API: GLM-5.3-Flash and MiMo-V2.6-Flash 🚀

Two more open models are live on the serverless API, next to Qwen3.8 27B. Same key, same endpoint, served from our own machines in Stockholm.

GLM-5.3-Flash (Z.ai). A model that thinks before every answer. Good for code, analysis and agents that have to reason through a task. Its thinking counts as output tokens. €0.35 per million input tokens, €2.16 per million output tokens. Model name glm-5.3-flash.

MiMo-V2.6-Flash (Xiaomi). 310B parameters with 12B active, so it answers fast and calls tools quickly. Built for agents. It thinks when you ask it to. €0.33 per million input tokens, €2.08 per million output tokens. Model name mimo-v2.6-flash.

Both have a 262,144-token context and read text only. Each one runs on two linked DGX Spark (GB10) machines. Prices are per million tokens, excl. VAT.

Try them

• GLM-5.3-Flash: https://glm-5-3-flash.axforge.ai/
• MiMo-V2.6-Flash: https://mimo-v2-6-flash.axforge.ai/
• Chat API docs: https://axforge.ai/docs/chat/

Already sending requests to Qwen? Change the model name and you are done. Tell me here what you build with them.