GLM-5.3 in a TEE.

• Private

  Open-weight models answer inside a trusted execution environment (TEE). Your browser verifies the enclave, then encrypts each message to it, so not even Thinkr can read the chat.

• ZDR

  Claude, GPT, Gemini, and other models answer with zero data retention (ZDR). The provider reads the chat to answer it and promises by contract to keep nothing.

In ZDR and Open, Thinkr's proxy can read the chat in transit and writes none of it to disk.

• GLM 5.3 FlashZ.ai's fast multimodal model for coding and long tasks.See the model
• GLM 5.3 FlashXA high-speed GLM 5.3 Flash, up to 200 tokens a second.See the model
• GLM 5.2Z.ai's earlier reasoning model for long agent workflows.See the model