大模型价格-性能曲线正在急剧右移，做模型选型和成本优化的团队可以直观看到哪些价位段性价比最高，建议点开看具体数据。
Text Arena 分析了自 2023 年以来大模型价格-性能帕累托前沿的 5 个模式。GPT-4 级别质量的成本从 2023 年的约 50 美元/百万 tokens 降至如今的约 0.10 美元，降幅达 500 倍。高端模型性能提升约 170 分（从 1330 到 1500），同时价格从约 50 美元降至约 20 美元。低价端（低于 0.20 美元）的模型性能从约 1000 分提升至约 1440 分，与顶级模型的差距从 350 分缩小到约 60 分。主要玩家轮换：OpenAI 奠定基准，Meta 加强低价端，Google DeepMind 推动 2025 年跃升，Anthropic 在 2026 年保持领先，xAI 和中国实验室（DeepSeek、零一万物、Kimi、小米、阿里通义）持续推动中端前沿。
5 patterns in Text Arena's price–performance Pareto frontier since 2023: 1. GPT-4-level quality i...
5 patterns in Text Arena's price–performance Pareto frontier since 2023: 1. GPT-4-level quality is now ~500x lower cost. - From a ~$50 blended price per million tokens in 2023 to ~$0.10 today. 2. The higher-price end is both better and lower-priced since 2023. - The leading Arena score has climbed ~170 points (1,330 → 1,500). While the price of the higher-end frontier models dropped from ~$50 to ~$20 per million tokens. 3. The low-cost end gained the most. - Under $0.20 per million tokens, the best available model went from ~1,000 Arena score in 2023 to ~1,440 today. 4. The low-cost/top performance gap has nearly closed. - In 2023, sub-$0.20 models trailed the leader by ~350 Arena points. Today, ~60. 5. The cast has rotated quite a bit. - - @OpenAI set the 2023–24 benchmark. - @AIatMeta strengthened the low-cost end in 2024. - @GoogleDeepMind drove the 2025 jump. - @AnthropicAI holds the peak in 2026. - @xAI and Chinese labs like @DeepSeekAI , @Zai_org , @Kimi_Moonshot , @XiaomiMiMo , and @Alibaba_Qwen are continuing to push the mid-price frontier. Your browser does not support the video tag. 🔗 View on Twitter 💬 1 🔄 0 ❤️ 3 👀 173 📊 1 ⚡