XIPU AI Bridge v0.5.0

Changes

• Support the current XIPU AI /chat frontend alongside /v3/chat, with setup guidance pointing to /chat.
• Accept PDF and UTF-8 text documents through Responses input_file and Chat Completions file parts. Preserve mixed attachment order and apply frontend limits: up to ten attachments on /chat, or one image on /v3/chat.
• Report failed attachments by position and filename, with machine-readable request locations, failure stages, and upstream error details.
• Improve Linux native-host registration for Chromium, Brave, and custom configuration directories.
• Clarify setup, settings, installation instructions, API compatibility limits, and debugging options.

Install or upgrade

Download both the native archive for your operating system and CPU and xipu-bridge-extension.zip. Extract both and follow the included INSTALL.md. In Chrome, enable Developer mode at chrome://extensions, choose Load unpacked, and select the extracted extension folder.

For upgrades, quit the browser before running the native installer. Replace the files in the existing unpacked extension folder, reopen the browser, reload the extension, and refresh the signed-in XIPU AI tab. Existing host settings and the local API key are preserved.

Prebuilt binaries are not Developer ID or Authenticode signed, or Apple notarized; the operating system may warn or block execution.

Compatibility

Document uploads require /chat. Upstream upload availability and document recognition vary; the attachment-count limit does not guarantee successful generation with every combination of files. Tool calls and structured output use prompt-based compatibility and local validation. Client token limits do not cap generation or charges, and per-request token usage remains unavailable.

See the reference and verification notes for supported behavior and platform testing limits.