How good are open-source Chinese models?
I’ve been using GitHub Copilot and I’m honestly impressed by its capabilities, especially for debugging and automating tasks.
For example, I was working on an app bug where I used 5.4 mini xhigh and it was still able to:
Open a browser and run Playwright scripts
Handle Excel file checks (like verifying downloads)
Debug frontend layout issues
Even take screenshots and suggest Flexbox fixes based on UI behavior
That got me curious, so I tried replicating similar workflows using some free/open-source models and tools. Using openrouter with GHCP with the free deepseek V4 flash model it wasn't able to do it
Now I’m wondering:
How good are the open-source Chinese models compared to tools like Copilot or GPT-based models?
People often say they’re strong in coding and agent-like workflows—how true is that in real-world use?
And can open code with the go subscription. Actually able to do something like this?