TypeSafe发布System One模型Jev，主打结构化决策与低延迟
Typesafe发布System One模型与Jev
提出System One新范式，用并行采样和结构化输出解决LLM自动化痛点，性能数据硬核，值得Agent开发者关注。
TypeSafe AI
TypeSafe AI
Manifesto
宣言
Our Team
我们的团队
Join Waitlist
加入候补名单
TypeSafe AI
TypeSafe AI
Manifesto
宣言
Our Team
我们的团队
∵ Back
∵ 返回
Company News
公司新闻
Sep 15, 2026
2026年9月15日
Introducing System One Models & Jev
推出 System One 模型与 Jev
Diogo Almeida, founder, TypeSafe
Diogo Almeida，TypeSafe 创始人
Models have been superhuman at chat for years, so where is all the automation?
多年来，模型在聊天方面已展现出超越人类的能力，那么所有的自动化在哪里呢？
This has been my driving question for the last four years. At OpenAI, I helped build the methods that made language models useful at following instructions and talking with people. That work ended up as the research behind ChatGPT. At the time, I thought maybe chat models would lead to AGI, but despite the hype it became obvious to me that there was something really big missing.
这是我过去四年来的核心驱动力。在 OpenAI，我帮助构建了使语言模型能够遵循指令并与人交流的方法。这些工作最终成为了 ChatGPT 背后的研究基础。当时，我曾认为聊天模型可能会通向 AGI（通用人工智能），但尽管 hype 不断，我却清楚地意识到，还缺少一些真正重要的东西。
After two years in stealth, countless technical challenges, and research breakthroughs… I am beyond excited to announce that today, TypeSafe AI is releasing our first System One Model: a new class of frontier models built to make fast, structured decisions that software can use directly.
经过两年的秘密研发、无数技术挑战以及研究突破……我非常兴奋地宣布，今天 TypeSafe AI 正式发布我们的首个 System One 模型：一类全新的前沿模型，旨在让软件能够直接利用其快速、结构化的决策。
We built a new stack entirely focused on automation: with a new model architecture, parallel sampler for maximum efficiency, and training method we call Reinforcement Learning for Calibrated Decisions (RLCD).
我们构建了一个完全专注于自动化的全新技术栈：包括新的模型架构、用于最大化效率的并行采样器，以及我们称之为“校准决策强化学习”（RLCD）的训练方法。
Our first public model is Jev, available today in early access. Jev achieves similar levels of intelligence on System One tasks compared to existing LLMs, while being two orders of magnitude faster and more efficient. While Jev gives up string generation, it’s optimized for structured outputs and can’t hallucinate.
我们的首个公开模型是 Jev，现已开放早期访问。Jev 在 System One 任务上达到了与现有大语言模型（LLM）相似的智能水平，同时速度更快、效率高出两个数量级。虽然 Jev 放弃了字符串生成能力，但它针对结构化输出进行了优化，且不会产生幻觉。
Think of Jev as a frontier-intelligence function call: unstructured state in, typed probabilistic decisions out.
可以将 Jev 视为一种前沿智能的功能调用：输入非结构化状态，输出类型化的概率决策。
Extraordinary claims require extraordinary evidence so see below for the receipts. 💅
非凡的主张需要非凡的证据，详情请见下文。💅
Frontiers, Old and New
新旧前沿
| Existing LLMs | System One + Jev | |
|---|---|---|
| Optimized with | Reinforcement Learning with Human Feedback (RLHF) / Reinforcement Learning with Verifiable Rewards (RLVR) | Reinforcement Learning for Calibrated Decisions (RLCD) |
| Optimizes for | Human preference: writeups and chat responses that human raters prefer.Verifiable rewards: outputs that can be programmatically verified. | Calibrated decisions: answers with epistemically honest probabilities on System One tasks. |
| Inputs | Unstructured data (e.g. text) with an emphasis on sequential messages. | Unstructured data (e.g. text) with an emphasis on structured program state. |
| Outputs | Strings / generated text. Strings are flexible and can be anything: chat responses, code, hallucinations, refusals, or even type-safe structured values. To be used by software, responses need to be parsed + validated. There is also always some risk that the AI goes off the rails. | Type-safe structured values. Possible outputs and structure are defined in advance. The model never makes type errors. All answers are accompanied with calibrated probabilities and confidence scores. |
| Sampling | Sequential. Generates one token at a time, each conditioned on the last. | Parallel. Generates all outputs in a single query. Incredibly efficient and hardware-aware. |
| Cost | Input tokens: from $0.20 to $10 / MTok.Output tokens: ~5x more expensive than input tokens. | Input tokens: $0.042 / MTok ($42 per billion tokens).Output tokens: FREE (too cheap to meter). |
| Speed | End-to-end response time is 3 to 329 seconds for frontier models. Fast enough for interfacing with humans, but a big bottleneck when integrated in code. | End-to-end response time is 70ms-500ms for TypeSafe. This can range from 40x-200x faster for the same levels of frontier intelligence for System One shaped queries. |
| Confidence | Even if prompted for a confidence estimate, models tend to be overconfident and inconsistent. If a model can do a task 95% of the time but doesn’t say when it’s in the 5%, it can’t automate that task. | Always communicates confidence and uncertainty with every output. Calibrated: higher confidence means higher accuracy. More consistent: returns similar answers for similar inputs. |
| Use cases | Human-in-the-loop tasks (chatbots, copilots, coding agents). General and powerful, but requires human oversight because their freedom also means they might go off the rails.Verifiable problems (math proofs, kernel optimization). When correctness can be checked cheaply and automatically, LLMs can generate, test, and iterate until they find something that works.Demos. The flexibility of strings allows it to be incredible for quickly making prototypes that only work sometimes. | AI-Powered Workflows / smart if-statements. Structured outputs slot into ordinary software as fuzzy decision rules: classify, route, score, extract, or branch where hand-written logic is too brittle. The surrounding code constrains their freedom, making them easier to compose into reliable systems.Map-reducing over big data. Turn petabytes of data into features and insights.Real-time applications. 100ms speeds means you can use AI in your applications where UX is critical.Verify everything. Score, judge, verify, guardrail, and detect jailbreaks of LLM prompts, reasoning traces, and/or outputs. |
| 现有大语言模型 | System One + Jev | |
|---|---|---|
| 优化方式 | 基于人类反馈的强化学习 (RLHF) / 基于可验证奖励的强化学习 (RLVR) | 用于校准决策的强化学习 (RLCD) |
| 优化目标 | 人类偏好：人类评估者更偏好的写作和聊天回复。可验证奖励：可通过程序验证的输出。 | 校准决策：针对 System One 任务提供认识论上诚实的概率答案。 |
| 输入 | 非结构化数据（如文本），强调顺序消息。 | 非结构化数据（如文本），强调结构化程序状态。 |
| 输出 | 字符串 / 生成的文本。字符串具有灵活性，可以是任何内容：聊天回复、代码、幻觉、拒绝响应，甚至是类型安全的结构化值。若要被软件使用，响应需要解析和验证。此外，AI 偏离正轨的风险始终存在。 | 类型安全的结构化值。可能的输出和结构预先定义。模型永远不会出现类型错误。所有答案均附带校准后的概率和置信度分数。 |
| 采样 | 顺序生成。一次生成一个 token，每个 token 都以前一个为条件。 | 并行生成。在单次查询中生成所有输出。效率极高且感知硬件特性。 |
| 成本 | 输入 token：每百万 token ($MTok) 从 $0.20 到 $10。输出 token：比输入 token 贵约 5 倍。 | 输入 token：$0.042 / MTok（每十亿 token $42）。输出 token：免费（便宜到无需计量）。 |
| 速度 | 端到端响应时间为 3 到 329 秒（前沿模型）。对于与人交互来说足够快，但在集成到代码中时是一个巨大的瓶颈。 | TypeSafe 的端到端响应时间为 70ms-500ms。对于相同水平的 System One 形状查询的前沿智能，速度可提高 40x-200x。 |
| 置信度 | 即使提示要求提供置信度估计，模型往往也过于自信且不一致。如果模型能完成某项任务 95% 的时间，但不说明那 5% 的情况，它就无法自动化该任务。 | 每次输出始终传达置信度和不确定性。校准良好：高置信度意味着高准确率。一致性更强：对相似输入返回相似答案。 |
| 用例 | 人在回路中的任务（聊天机器人、副驾驶、编码代理）。通用且强大，但需要人工监督，因为它们的自由度也意味着可能会偏离正轨。可验证问题（数学证明、内核优化）。当正确性可以廉价且自动检查时，LLM 可以生成、测试并迭代，直到找到可行的方案。演示原型。字符串的灵活性使其非常适合快速制作仅偶尔可用的原型。 | AI 驱动的工作流 / 智能 if 语句。结构化输出作为模糊决策规则融入普通软件：分类、路由、评分、提取或分支，而手写逻辑在此处过于脆弱。周围代码限制了其自由度，使其更容易组合成可靠系统。大规模数据的 Map-reduce。将 PB 级数据转化为特征和洞察。实时应用。100ms 的速度意味着你可以在用户体验至关重要的应用程序中使用 AI。全面验证。对 LLM 提示词、推理轨迹和/或输出进行评分、评判、验证、护栏保护和检测越狱行为。 |
Evidence / Technical Results
证据 / 技术结果
We love skeptics, and are skeptics ourselves.
我们热爱怀疑论者，我们自己也是怀疑论者。
There are some claims you can easily verify:
有些主张你可以轻松验证：
- Speed per call: We truly are that fast, though our published evals are generally run from our laptops on the West Coast (this is where our service is currently based).
- Cost per call: We make our pricing transparent. We can’t prove it isn’t subsidized; we’ll need the long-term to prove the sustainability of our pricing (which we expect to go down, not up).
- No type errors: This would be an easy thing to falsify with just a single counter-example, but it is mathematically impossible.
- 每次调用的速度：我们确实非常快，尽管我们发布的评估通常是在西海岸的笔记本电脑上运行的（这也是我们服务目前的所在地）。
- 每次调用的成本：我们公开定价。我们无法证明其未受补贴；我们需要时间来证明我们定价的可持续性（我们预计价格会下降，而不是上升）。
- 无类型错误：这只需一个反例就很容易被证伪，但在数学上是不可能的。
For our bolder claims, we want to provide as much nuance as we can.
对于更大胆的主张，我们希望提供尽可能多的细微差别说明。
Side-by-side demonstration
并排演示
Our side-by-side demo shows a key difference between our models and LLMs: Jev outputs all probabilities in parallel instead of autoregressively generating by token. Strings are extremely powerful and general, but costly. “Giving up” strings actually gives us a lot of superpowers!
我们的并排演示展示了我们的模型与 LLM 之间的一个关键区别：Jev 并行输出所有概率，而不是通过 token 自回归生成。字符串极其强大且通用，但成本高昂。“放弃”字符串实际上给了我们许多超能力！
Nuance
细微差别
- For people with early access to TypeSafe, here is the actual query.
- The query is highly simplified and questions were chosen to have descriptive, human-readable keys so that the output on the screen is understandable.
- The state is also a short, dense, and detailed paragraph, to emphasize the difference in sampling methodology. The relatively shorter input paints our model in an advantageous light.
- For the keen eyed, for the recorded run, the only disagreement with GPT-5.6 Terra is on “Churn likelihood level”. The actual answer seems genuinely ambiguous to us.
- We used GPT-5.6 Terra with default reasoning for this example, because we’ve found it to be the most comparable at intelligence to Jev on average.
- Fun fact: a similar demo was what convinced us to go all-in in the direction of System One Models!
- 对于早期获得 TypeSafe 访问权限的人，这是实际的查询。
- 该查询高度简化，问题被选择具有描述性、人类可读的键，以便屏幕上的输出易于理解。
- 状态也是一个简短、密集且详细的段落，以强调采样方法的差异。相对较短的输入使我们的模型看起来更有优势。
- 对于眼尖的人，在录制的运行中，与 GPT-5.6 Terra 的唯一分歧在于“流失可能性水平”。对我们来说，实际答案似乎确实是模棱两可的。
- 在这个例子中，我们使用了带有默认推理的 GPT-5.6 Terra，因为我们发现它在智力上与 Jev 平均最可比。
- 有趣的事实：类似的演示正是促使我们全力投入 System One Models 方向的原因！
Workflow evals
工作流评估
We made a new type of evaluation to measure how well AI works within code. We don’t optimize for a ground truth classification orand allow the harness and model to change (potentially allowing for overfitting via harness engineering). Instead, we assume there is a correct compute graph (a “workflow” represented in code) and use the predictions of the largest, smartest, and most expensive external models as reference probabilities.
我们创造了一种新的评估类型，用于衡量 AI 在代码内的工作表现。我们不针对真实分类进行优化，也不允许测试框架和模型发生变化（这可能通过框架工程导致过拟合）。相反，我们假设存在一个正确的计算图（用代码表示的“工作流”），并使用最大、最聪明和最昂贵的外部模型的预测作为参考概率。
Rephrased: every model gets the same workflow. We test how they compare to the average of the smartest models (in this case, Astra and Fable).
换句话说：每个模型都获得相同的工作流。我们测试它们与最聪明模型的平均水平（在本例中为 Astra 和 Fable）的比较情况。
Jev is off the charts – owning the Pareto frontier for almost 2 orders of magnitude. We also compare to models with a generated prompt doing all the logic in their chain-of-thought, but this tends to do significantly worse than using the workflow itself.
Jev 的表现远超预期——在接近两个数量级的范围内占据了帕累托前沿。我们还与那些在思维链中生成提示词并执行所有逻辑的模型进行了比较，但这类模型的表现往往显著低于直接使用工作流本身。
Note that the calls here are significantly more complex than the side-by-side demonstration above. That’s because they’re more representative of the types of production workloads needed for true business automation. Below is the simplest of the 4 workflows we’re publishing:
请注意，此处的调用比上述并列演示要复杂得多。这是因为它们更能代表实现真正业务自动化所需的典型生产负载。以下是我们要发布的四种工作流中最简单的一种：
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力