Pokee AI 发布 Isaac 28B：千万 token 上下文，可本地部署
Pokee AI Releases Pokee-Isaac 28B: A 10M-Token Context Agentic Model Built to Run Inside the Customer Boundary
Long-horizon agents accumulate context faster than they resolve tasks. Every tool output, observation, and intermediate reasoning step stays in the window, and the two capabilities that matter — holding that context and staying coherent across it — have so far been available almost exclusively from cloud endpoints. That excludes regulated industries, public-sector institutions, and on-device applications, where the data is not permitted to leave the boundary at all. Pokee AI released Pokee-Isaac 28B, a 28B text-only foundation model with a 10M-token context window, designed to run inside that boundary. The Pokee research team claims 93.3% on RULER at 10M tokens, parity with the strongest cost-optimized cloud baselines on agentic benchmarks, and a serving profile that fits a single GPU.
长时程智能体积累上下文的速度快于它们解决任务的速度。每一个工具输出、观察结果和中间推理步骤都保留在窗口中，而两项关键能力——保持该上下文并在其中保持连贯——迄今为止几乎只能通过云端端点获得。这排除了受监管行业、公共部门机构和设备端应用，这些场景中数据完全不允许离开边界。Pokee AI 发布了 Pokee-Isaac 28B，一个拥有 1000 万 token 上下文窗口的 28B 纯文本基础模型，旨在该边界内运行。Pokee 研究团队声称在 10M token 下 RULER 得分为 93.3%，在智能体基准测试中与最强的成本优化云基线持平，并且其服务配置可适配单块 GPU。
Is it deployable
是否可部署
Yes — but licensed, not open-weight. Pokee AI serves Isaac through an OpenAI-compatible developer API, and licenses it for deployment inside a VPC, on-premises, or on-device. The launch announcement advertises Day-0 support for vLLM and SGLang, and single-GPU serving starting from an RTX 4090 or equivalent. The research team publishes measurements only from a single B200-class GPU, so treat the consumer-GPU claim as vendor guidance rather than a reported result.
可以——但需获得许可，而非开放权重。Pokee AI 通过兼容 OpenAI 的开发者 API 提供 Isaac 服务，并授权在 VPC 内、本地或设备端部署。发布公告宣传对 vLLM 和 SGLang 的 Day-0 支持，以及从 RTX 4090 或同等 GPU 开始的单 GPU 服务。研究团队仅发布了来自单个 B200 级 GPU 的测量结果，因此请将消费级 GPU 的声明视为供应商指导，而非已报告的结果。
- Company level: This fits organizations that already own their inference stack — mid-size and enterprise teams with a platform group, plus device OEMs. A solo practitioner without on-prem hardware should use the hosted API instead; the boundary argument only pays off if you have a boundary.
- Industries: Healthcare and payors, financial services and insurance, defense and public sector, legal and e-discovery, and pharma or semiconductor R&D. The common trait is a rule that the data cannot cross an external API boundary, not a preference for privacy.
- Applications: Whole-repository code review, multi-year contract and claims analysis, incident forensics over full log archives, and long-running tool agents that never need summarization or context pruning. The research paper makes this second point explicitly: when enough usable context is available in-boundary, memory hierarchies and compression become optional rather than required.
- 公司层面：这适合已经拥有自有推理栈的组织——拥有平台团队的中型和企业级团队，以及设备 OEM。没有本地硬件的独立从业者应改用托管 API；边界论证只有在你有边界时才有效。
- 行业：医疗保健和支付方、金融服务和保险、国防和公共部门、法律和电子发现，以及制药或半导体研发。共同特征是数据不能跨越外部 API 边界的规则，而非对隐私的偏好。
- 应用：整个代码库的代码审查、多年期合同和索赔分析、完整日志档案的事件取证，以及无需摘要或上下文修剪的长期运行工具智能体。研究论文明确指出了第二点：当边界内有足够的可用上下文时，内存层次结构和压缩变得可选而非必需。
Long-context results
长上下文结果
On RULER, Isaac stays above 93.3% at every tested length, ending at 93.3% at 10M. GPT-5.6 Luna and Gemini 3.5 Flash Lite track it to 512K, then hit context-overflow at 1M.
在 RULER 上，Isaac 在每个测试长度下都保持在 93.3% 以上，在 10M 时最终为 93.3%。GPT-5.6 Luna 和 Gemini 3.5 Flash Lite 在 512K 之前与其持平，然后在 1M 时遇到上下文溢出。
On MRCR v2 with 8 needles, Isaac scores 0.607, 0.743, and 0.500 at 256K, 512K, and 1M. Its margin over Gemini widens from 0.133 to 0.295 across that sweep.
在配备8个针头的MRCR v2上，Isaac在256K、512K和1M上下文长度下的得分分别为0.607、0.743和0.500。其相对于Gemini的优势从0.133扩大到0.295。
Agentic and security results
代理与安全结果
Isaac leads BFCL v4 at 70.94 against Luna’s 70.61. The report calls that parity rather than a lead, which is the correct read. On τ³-bench it averages 0.662 across four domains, ahead of Gemini’s 0.631, with banking at 0.186 for everyone’s difficulty. On MCP-Atlas it places third at 74.59% coverage, but uses 9.10 turns per task against Gemini’s 14.99. On Terminal-Bench 2.1 it resolves 56 of 86 text-compatible tasks (65.1%), behind Luna’s 60. That is the one benchmark a cloud baseline wins, and the report states it plainly.
Isaac在BFCL v4上以70.94领先，而Luna为70.61。报告称这是持平而非领先，这是正确的解读。在τ³-bench上，它在四个领域的平均分为0.662，领先于Gemini的0.631，其中银行领域为0.186，对所有人来说都颇具难度。在MCP-Atlas上，它以74.59%的覆盖率排名第三，但每个任务使用9.10轮，而Gemini为14.99轮。在Terminal-Bench 2.1上，它解决了86个文本兼容任务中的56个（65.1%），落后于Luna的60个。这是云基线唯一获胜的基准，报告明确指出了这一点。
On DTAP red-teaming, Isaac records the lowest direct (36.0), indirect (35.2), and combined (35.6) attack success rates, with 82.5 benign success. One condition differs: baselines ran under the stock runner, Isaac under the Pokee harness.
在DTAP红队测试中，Isaac记录了最低的直接（36.0）、间接（35.2）和综合（35.6）攻击成功率，良性成功率为82.5。一个条件不同：基线在标准runner下运行，而Isaac在Pokee harness下运行。
Efficiency, pricing, and portability
效率、定价与可移植性
Under the RULER workload on one B200-class GPU, TTFT is 23.6s at 1M and 72.9s at 10M. Prefill throughput rises with context, from 42,400 to 137,200 tokens/s, so a ten-fold longer prompt costs roughly three times the TTFT. List pricing is $0.15/$1.00 per million input/output tokens, marked provisional. Isaac also runs fully on-device on Intel Arc Pro B70 and Core Ultra Series 3 (Panther Lake), and on Qualcomm Snapdragon X2 Elite.
在单块B200级GPU上的RULER工作负载下，TTFT在1M时为23.6秒，在10M时为72.9秒。预填充吞吐量随上下文增加，从42,400 tokens/s上升到137,200 tokens/s，因此十倍的提示长度大约需要三倍的TTFT。标价暂定为每百万输入/输出token $0.15/$1.00。Isaac还可以完全在设备端运行，支持Intel Arc Pro B70和Core Ultra Series 3（Panther Lake），以及Qualcomm Snapdragon X2 Elite。
Key Takeaways
关键要点
- Pokee-Isaac 28B scores 93.3% on RULER at 10M tokens; every baseline in its panel returns 0.0 beyond 2M.
- Prefill reaches 137,200 tokens/s at 10M context on one B200; decode holds flat near 335 tokens/s.
- It leads BFCL v4 (70.94) and τ³-bench (0.662 avg), places second on Terminal-Bench 2.1, third on MCP-Atlas.
- Lowest combined attack success rate on DTAP (35.6) while keeping 82.5 benign task success.
- Weights are not published; deployment is licensed into VPC, on-premises, or on-device.
- Pokee-Isaac 28B在10M token的RULER上得分93.3%；其面板中的每个基线在超过2M时都返回0.0。
- 在单块B200上，10M上下文时预填充达到137,200 tokens/s；解码稳定在约335 tokens/s。
- 它在BFCL v4（70.94）和τ³-bench（平均0.662）上领先，在Terminal-Bench 2.1上排名第二，在MCP-Atlas上排名第三。
- 在DTAP上综合攻击成功率最低（35.6），同时保持82.5的良性任务成功率。
- 权重未公开；部署需许可到VPC、本地或设备端。
Check out the Blog and Paper. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
查看博客和论文。另外，欢迎在Twitter上关注我们，别忘了加入我们的150k+ ML SubReddit并订阅我们的新闻通讯。等等！你在Telegram上吗？现在你也可以在Telegram上加入我们。
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
需要与我们合作推广你的GitHub仓库、Hugging Face页面、产品发布或网络研讨会等？请联系我们。
The post Pokee AI Releases Pokee-Isaac 28B: A 10M-Token Context Agentic Model Built to Run Inside the Customer Boundary appeared first on MarkTechPost.
文章《Pokee AI Releases Pokee-Isaac 28B: A 10M-Token Context Agentic Model Built to Run Inside the Customer Boundary》首次出现在MarkTechPost上。
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