ModelsSelected

Thinky Machines 的 Inkling 2 Small 即将开源,有望登顶美国开源模型

Why we picked it

Thinky Machines 把 Inkling 2 Small 的成绩拉到 Agent Arena 第 31 位,比上一代涨了约 18 个百分点,成本只要 Kimi K3 (Max) 的三成左右。开源权重一出,它就是美国最强的开源模型,做 Agent 和微调的开发者可以直接在 Tinker 上先试起来。

Thinky Machines 宣布即将开放 Inkling 2 Small 的模型权重。在 Agent Arena 排行榜上,Inkling 2 Small 目前总排名第 31 位,净提升 +0.08%,较上一代 Inkling-Small(第 56 位、-18%)提升了约 18 个百分点。权重发布后,它将成为排名最高的美国开源权重模型,在所有开源模型中位列第 9。

Body

与头部开源模型相比,Inkling 2 Small 距离第 5 位的 Kimi K3 (Max) 仅差 0.5 个百分点,距离第 3 位的 DeepSeek-V4.1-Flash-20261017 (Max) 在 2.0 个百分点以内。成本方面,其每任务中位成本为 $0.25,比 Kimi K3 (Max) 低 70%,处于成本-性能 Pareto 前沿附近。

除了 Agent Arena,Inkling 2 Small 以 1648 分在 Code Arena: WebDev 的实验室排名中列第 8。模型目前已可在 Tinker 平台上进行微调和推理,完整权重文件即将放出。

Original · lmarena.ai

Exciting news: With open weights coming soon, Inkling 2 Small by @thinkymachines is set to become the #1 US open-weights model in Agent Arena! Inkling 2 Small landed #31 overall with +0.08% net improvement, up about 18 percentage points from Inkling-Small at #56 with −18%. Among all open models, it would rank #9 , just 0.5 percentage pts from Kimi K3 (Max) at #5 and within 2.0 pts of DeepSeek-V4.1-Flash-20261017 (Max) at #3 . Inkling 2 Small’s median cost per task is $0.25, 70% lower than Kimi K3 (Max). See its position relative to the Pareto frontier below. Congrats to the @thinkymachines team on this release! Arena.ai @arena ICYMI: Once its weights are released, Inkling 2 Small from @thinkymachines would be the #1 US open model. With 1648 pts, it’s #8 overall among labs in the Code Arena: WebDev! Inkling 2 Small is available now for fine-tuning and inference on Tinker, with full weights coming soon. Congrats again to the @thinkymachines team! 🔗 View Quoted Tweet📊 13 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Inkling 2 Small, Thinky Machines, 开源模型, 智能体.

Source attribution: curated from lmarena.ai (@lmarena_ai).