DeepSeek 又发新模型了，V4-Pro (Max) 在编码榜上排开源第二，比贵得多的 Opus-4.8 还强，价格只要零头。
DeepSeek-V4-Pro (Max) 在 Code Arena: WebDev 的 AutoEval 中获得 1607 分，总分第 8，开放权重模型中排第 2。该成绩仅次于 GPT-5.6 Sol 的 1622 分和 Kimi K3 (Max) 的 1674 分。其定价为每百万 tokens 输入 $0.435、输出 $0.87，低于 Opus-4.8 的 $5/$25 和 GLM-5.2 的 $1.4/$4.4。目前该分数为早期 AutoEval 结果，后续会随真人投票收敛。
DeepSeek-V4-Pro (Max) by @deepseek_ai is expected to shift the Pareto curve for Code Arena: WebDev w...
DeepSeek-V4-Pro (Max) by @deepseek_ai is expected to shift the Pareto curve for Code Arena: WebDev with this upcoming open weights model. It currently sits at ~ #8 overall (AutoEval) at 1607 pts. Priced at $0.435 input/ $0.87 output per MTokens, it outperforms models beyond its price tier, including Opus-4.8 ($5/$25) and GLM-5.2 ($1.4/$4.4). Congrats again to the @deepseek_ai team for their contribution to the open ecosystem! Arena.ai @arena Big news: DeepSeek-V4-Pro (Max) by @deepseek_ai is coming in around ~ #8 overall ( #2 among open models) in the Code Arena: WebDev! At 1607 pts, this places it after GPT-5.6 Sol (xHigh)(1622 pts), and makes it the second best open model after Kimi K3 (Max) (1674 pts). Note: this is an early AutoEval score, in which a Reward Model trained on Arena's human preference data casts automatic votes in place of live votes. We’ll continue to see how scores converge as more live human votes come in. See thread for Text Arena scores and more on AutoEval's methodology. Congrats to the @deepseek_ai team on this release! 🔗 View Quoted Tweet 💬 0 🔄 0 ❤️ 10 👀 1556 📊 1 ⚡