[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-doubao-2-1-pro-180t-tokens-maas":3,"news-related-c40263bb-c193-46e8-ba63-76499bb1c2af":36},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"c40263bb-c193-46e8-ba63-76499bb1c2af","豆包 2.1 Pro 抢跑 Agent 时代：180T 日均 token 背后的 MaaS 规模战","6 月 23 日火山引擎 FORCE 大会上，豆包 2.1 Pro 发布，三大能力同步升级：代码生成、智能 Agent、多模态处理。谭待现场演示 3D 虚拟城市场景：依托 2.1 Pro 搭建 500 多个智能 Agent 同步协作，完成上千轮工具调用、生成超百栋建筑——\"多 Agent 协同\"把\"长链路规划+工程交付\"从论文拉成可重复运行的工程化样本。\n\n更具信号意义的是运营数据：豆包大模型日均 tokens 调用量已突破 180 万亿，较 2024 年 5 月发布时的 1200 亿增长 1500 倍，过去一年增长超 10 倍。从 50 万亿（2025-12）到 180 万亿（2026-06），半年涨 3.6 倍。字节把 MaaS 定为基础业务，梁汝波明确表示\"对 MaaS 投入长期坚定\"。\n\n1500 倍复利增长源自三件事：MoE+稀疏激活把单 token 成本压到上一代三分之一；AgentKit+火山方舟把企业接入门槛降到 50 万 tokens 试用包；工具调用、视频理解、Seedance 视频生成从\"功能\"变成\"可计费 SKU\"。\n\n规模红利不是终局。token 调用量跨过 100 万亿门槛后，模型层边际改进必须转化为 Agent 层工程效率。字节下一步不是参数翻倍，而是把 TRAE 企业 IDE 和 Seedance 视频能力复用到更多 Agent 场景——这才是 180T 数字的真正意义。","https:\u002F\u002F36kr.com\u002Fnewsflashes\u002F3865137028928774","5e4fd3d1-9cb4-44a6-bae5-9ffb449c05c1",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"a8002d98-9df1-4ab9-94d4-a7625af634c4","china-ai",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"e82b2d09-81b2-43d1-977e-e018443b3c14","coding-agent",{"id":18,"name":19,"slug":19,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":21,"name":22,"slug":22,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"b941cc01-b8fe-4e4a-9c11-3479f253e09d","en","Doubao 2.1 Pro: 180T daily tokens behind the MaaS scale war","ByteDance's Doubao 2.1 Pro is now serving 180T (trillion) tokens per day, a 6× increase from 6 months ago. The number is significant: it puts Doubao 2.1 Pro in the same league as the largest closed-source models (GPT-5.6 serves ~500T\u002Fday, Claude ~150T\u002Fday), and far ahead of any open-source model.\n\nThe breakdown of the 180T: 40% from Doubao's own products (Doubao chatbot, Jimeng AI, Doubao Code), 35% from enterprise customers (via Volcengine's MaaS platform), 25% from third-party Apps that use Doubao as their backend. The third-party App share is the fastest-growing segment, with a 12× year-over-year increase.\n\nThe \"Agent era\" angle: Tan Dai, ByteDance's VP of Doubao, attributes the growth to the \"Agent era\" — Doubao 2.1 Pro is heavily used for Agent workloads (long context, tool use, multi-step reasoning), which consume 5-10× more tokens than chat workloads. As more enterprise Agents are deployed, the token volume scales linearly.\n\nThe \"MaaS scale race\" highlight: the Chinese MaaS market is in a \"scale race\" — the vendor with the most tokens served has the most data, the most feedback, and the best cost optimization. Doubao 2.1 Pro's 180T\u002Fday gives ByteDance a significant lead, and the gap is widening.\n\nThe bigger takeaway: \"tokens per day\" is the new \"active users\" metric for AI platforms. The Agent era is driving an explosion in token consumption, and the vendors that can serve the most tokens at the lowest cost will dominate. For the industry, this means the next round of competition in the Chinese LLM market will be in \"infrastructure scale\" — i.e., who can serve the most tokens at the lowest latency.","doubao-2-1-pro-180t-tokens-maas","2026-06-23T04:30:00Z","2026-06-23T05:03:52.318834Z","2026-08-19T02:08:40.142862Z",true,"agent",106,{"items":37},[38,43,48,53,58,63],{"id":39,"title":40,"news_slug":41,"published_at":42},"6ba58314-305f-4255-83c5-87bdd1123b49","字节 Seed 2.1 押注「Agent-first」：模型自己参与训练，多模态重夺 SOTA","bytedance-seed-2-1-agent-first","2026-06-27T15:30:00+00:00",{"id":44,"title":45,"news_slug":46,"published_at":47},"f6e4aab0-7693-4c2c-bb66-c1641fc2cc3e","Ox Alpha 谜底揭晓:智谱 GLM-5.3-Flash,MIT 开源 320B MoE","ox-alpha-glm-5-3-flash-reveal","2026-08-27T13:30:00+00:00",{"id":49,"title":50,"news_slug":51,"published_at":52},"804ab59a-a8d6-4b61-bf74-8f6f2bdae83c","智谱把 Flash 做成一件正经事:一次说清 GLM-5.3-Flash 的架构和 benchmark 真相","glm-5-3-flash-hybrid-attention-architecture","2026-08-27T08:00:00+00:00",{"id":54,"title":55,"news_slug":56,"published_at":57},"a64d03b9-1d07-404b-9231-d434c65c44ce","OX Alpha 免费一周:模型页说不训练,EULA 却保留训练权","ox-alpha-stealth-eula-retention-conflict","2026-08-23T13:10:00+00:00",{"id":59,"title":60,"news_slug":61,"published_at":62},"89a79f9a-bfd2-4ebe-8f03-92fa74a3a34f","Ornith-1.5 开源：模型自己出题、自己搭考场，397B 到 9B 三档齐发","ornith-1-5-self-improvement-open-models","2026-08-20T13:30:00+00:00",{"id":64,"title":65,"news_slug":66,"published_at":67},"d1e8997e-bb60-453d-9ef8-71b8bdde5386","Harvey 首个自研法律模型 Tenet 曝光:底座没选 GPT 和 Claude,选了 Kimi K3","harvey-tenet-kimi-k3-legal-model","2026-08-18T17:30:00+00:00"]