你签了另一个标签或窗口。 重新装入以刷新您的会话。 你在另一个标签或窗口里签字了重新装入以刷新您的会话。 您在另一个标签或窗口上切换账户。 重新装入以刷新您的会话。 解除警报
GVS5H:5个Qwen3.8-27B 在 LiveCode Bench 硬体上匹配 Claude 功能 5的模型
GVS5H：大猩猩对5人
结果
Opus 5运行在更早的版本的脚手架上，单传通过OpenRouter服务，而不是5传在固定的后端。
摘要。 边疆编码性能一般以更大的专有模式购买，成本高昂。 我们采用基于分类账的零镜头自控方法，这是一种无训练方法，在一个模型的新实例中，通过持有计划，指出和当前解决方案的共享文件系统来分解问题和协调。 在100个最新硬LiveCode Bench问题的11个开放型和封闭型模型中，该方法在固定后端产生25.6个百分点的增益，并提供两条通往前沿准确度的路线。 管弦乐Qwen3.8 Flash Next达到93.0% pass@1对Fable 5的90.4%，成本为9%，本地服务，开放重量的Qwen3.8-27B从66.8%上升到92.4%. 收益并不普遍：有些模型没有改变或更糟。 转写分析将收益归结为分解性和持久性上下文。 推论时间组织可以以成本的一小部分接近前沿编码精度，或者在可自我接受的权重上略微超过。
--engine多代理运行管理器；-engine单机是单调基准。
其他型号：anthropic:<model>，破损镜：<model>，开路器：<model>，每个型号都有自己的QQPI KEY.
传球@ 1 得分印在结尾。 记录结果为运行/结果.json，工作空间为运行/ws/.
许可证
密码是麻省理工学院的许可论文、数字和运行数据
CC BY 4.0 (英语). LiveCode Bench 叉，基准测试问题语句
和 LaTeX 模板文件保存自己的许可证。 见Notice.md
哪个许可证覆盖了哪条路。
关于
GVS5H:5个Qwen3.8-27B 在LiveCode Bench Hard 上匹配Claude Fable 5的型号 5 级编码，价格为第五，或者在单一的GPU上
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
GVS5H: Five Qwen3.8-27B Models Match Claude Fable 5 on LiveCodeBench Hard
GVS5H: Gorilla vs. 5 Humans
Results
Opus 5 ran on an earlier version of the scaffold, in a single pass served through OpenRouter rather than five on a pinned backend.
Abstract. Frontier coding performance is typically bought with larger proprietary models at high cost. We introduce ledger-based zero-shot self-orchestration, a training-free method in which fresh instances of one model decompose problems and coordinate through a shared filesystem holding a plan, notes and current solution. Across eleven open and closed-weight models on the 100 latest hard LiveCodeBench problems, the method yields gains of up to 25.6 percentage points on pinned backends and offers two routes to frontier-level accuracy. Orchestrated Qwen3.8 Flash Next reaches 93.0% pass@1 against Fable 5's 90.4% at 9% of the cost, and locally served, open-weight Qwen3.8-27B rises from 66.8% to 92.4%. Gains are not universal: some models are unchanged or worse. Transcript analysis attributes the gain to decomposition and persistent context. Inference-time organization can approach frontier coding accuracy at a fraction of the cost, or slightly exceed it on self-hostable weights.
--engine multiagent runs the manager; --engine single is the one-call baseline.
Other models: anthropic:<model>, dashscope:<model>, openrouter:<model>, each with its own *_API_KEY.
The pass@1 score prints at the end. Results are written to runs/results.json, workspaces to runs/ws/.
License
Code is under the MIT License. The paper, figures and run data are under
CC BY 4.0. The LiveCodeBench fork, the benchmark problem statements
and the LaTeX template files keep their own licenses. See NOTICE.md for
which license covers which path.
About
GVS5H: Five Qwen3.8-27B Models Match Claude Fable 5 on LiveCodeBench Hard | Fable 5 Level Coding for a Fifth the Price - or on a Single GPU
Auto-fetched and translated by AI Atlas. All rights belong to the original author.