⚡Megatron LLM Hub
←Back to search
Twitter

受邀测试者称 Qwen 4 在人工评测上超过 Qwen3.8 Max。

llm模型
RT @vysecurity: Our invite access to Qwen 4 (no idea what size) is beating Qwen3.8 Max in human eval, but mostly similar in other benchmarks. If this is a small model it’s beating even the big models.