I was ready to rage, but then I tried on GPT, Gemini, Claude, Copilot, Kimi, DeepSeek, GLM-5.3 and they all picked 17...
FML 😂 This is straight out of Doctor Who's "Extremis"
DeepSeek Harness vs Claude Code.
Same prompt, same website, same game. DeepSeek: 483,000 tokens. 11 minutes. Claude Code: 48,000 tokens. Still running 30 minutes later. DeepSeek finished faster and way cheaper. Claude built something more polished. Claude still has the edge.
Testing DeepSeek, OMP and OpenCode harnesses using the same Ox Alpha model.
I gave all 3 the same Ranger RF-31D reference image from Interstellar and the exact same one-shot prompt to see how well each harness could recreate it. Here’s the result 👇
There is something they often notice about NVIDIA
You can try out a large number of AI models via a free API from It includes examples of: Qwen Llama Gemma DeepSeek And others Choose the model Generate API Key And use it directly The interface is compatible with the OpenAI API This means that you can often install it easily in the existing tools…
You have reached the end of the archive
All of deepseek