Category: Performance Benchmarks

• Qwen3.5-Flash Arrives! Three Medium-Scale Models Open-Sourced
• Zhipu's New Model Also Uses DeepSeek's MLA, Runs on Apple M5
• OpenAI's Strongest Reasoning Model o3-pro Just Born! Crushing Gemini 2.5 Pro!
• DeepSeek makes a big move! New model focuses on mathematical theorem proving, significantly refreshing multiple high-difficulty benchmarks.