
OpenAI 直播活動
OpenAI 將舉辦一場直播活動。在直播期間將揭露具體的公告、產品發布或示範內容。
上一次 OpenAI 突然搞直播,他們直接丟出 GPT-4 Turbo,然後一夜之間改掉所有定價

Open-source AI is evolving insanely fast, but it’s hard to know which model is actually best for each use case. So I put together a list of the best open-source models across different categories Best Audio Generation Open Source Models Text-to-Speech (TTS) Qwen3-TTS → Best overall balance (quality + speed) Kimi-Audio → Strong multimodal + expressive voices Fish Speech / Fish Audio S2 → Great for realistic voice cloning CosyVoice 3.0 → Very solid multilingual + streaming VibeVoice Realt