Skip to content

[Bug] OpenCode provider responses are significantly slower in PawWork compared to CLI and other apps #1548

Description

@asaprajx

What happened?

I've noticed that using OpenCode as a provider inside PawWork is noticeably sluggish. Models take a long time to respond, both when using the built-in API and my own configured API keys. Interestingly, the exact same API endpoints run lightning-fast when I use the OpenCode CLI, the official OpenCode desktop app, or other BYOK apps.
(I wasn't able to grab a clean screenshot or screen recording of this since it's a latency/performance issue rather than a hard crash. It's totally possible this is just happening on my end, but I’ve tried switching network providers and troubleshooting locally, and the delay only happens inside PawWork, everything else flies.)

Which area seems affected?

Not sure

How much does this affect you?

Minor issue or polish

Steps to reproduce

1.Configure OpenCode API in PawWork (either built-in or custom key).
2.Send a prompt or query in a chat session and observe the time-to-first-token and overall generation speed.
3.Compare the same prompt/API key using the OpenCode CLI or another BYOK app to see the speed difference.

What did you expect to happen?

OpenCode API responses and model execution speeds in PawWork should match the snappy performance seen in the OpenCode CLI and other third-party clients.

PawWork version

v2026.8.1

OS version

Windows 11

Can you reproduce it again?

Sometimes

Diagnostics

It has stared happening after I updated to the first release of August.
The model takes long gaps between executing tasks

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions