CAI hangs when using Anthropic provided Claude model

[biggmojo]

Hello!

I'm trying to use cai for the first time and it seems to just hang when I attempt to use a Claude model provided by Anthropic? I started cai, chose the claude-sonnet-4-5 model, chose the redteam_agent agent, and asked for it to check what services are running on a machine on my local network. The log file was generated, then nothing happened for over 10 minutes before I exited out.

Here's a screenshot showing it:

And here's the content of my .env file:

OPENAI_API_KEY="empty"
ANTHROPIC_API_KEY="{redacted but is a valid key from Anthropic}"
OLLAMA=""
PROMPT_TOOLKIT_NO_CPR=1
CAI_STREAM=true

No errors or anything are thrown either when I Control+C out of cai. I tried following along with this YouTube demo to see if there was something I was missing or doing wrong but it didn't seem like it.

Other details that might be beneficial:

• Running cai in a Fedora VM.
• Installed with:

python3.13 -m venv cai_env
source cai_env/bin/activate
pip install cai-framework

• If I try to use a Claude model provided by Openrouter, running a prompt throws an error because I haven't authenticated to Openrouter or provided an access token (as expected).

I'm not really sure what else to try. Any suggestions? More than happy to provide additional information as necessary.

[0xhsn]

Same here, I ended up falling back to using OpenAI models instead. I’m running CAI in a container for reference.