| 1 | ai | 1 | 49 | Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU |
| 1 | ai | 1 | 47 | DeepSeek Nears OpenAI’s Cyber Model in Vals Test at $3 a Run |
| 1 | ai | 1 | 51 | DeepSeek’s Share of OpenRouter Top-10 Tokens Rises to 56% |
| 1 | ai | 1 | 39 | Free Chinese AI Models Are Getting Scary Good at Hacking |