Thinking Machines Lab 推出了 Inkling-Small，这是一款新的开源多模态模型，它优先考虑效率而非单纯的规模。尽管比其前身 Inkling 体积小得多，Inkling-Small 在各种编码和推理基准测试中表现出卓越的性能。该模型设计用于更轻松的部署，能够在一台 GPU 上运行，并根据 Apache 2.0 许可证提供。 AI
影响 此次发布预示着向更高效、可部署且仍能实现最先进性能的模型转变的潜力，从而降低了采用门槛。
排序理由 发布了具有系统卡和基准测试结果的前沿实验室模型。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
<p><img alt="" class="attachment-full size-full wp-post-image" height="801" src="https://the-decoder.com/wp-content/uploads/2026/07/thinking_machines_inkling_logo.png" style="height: auto; margin-bottom: 10px;" width="1256" /></p> <p> Thinking Machines, the AI lab from former Ope…
<p>Inkling-Small matches Inkling at a quarter the size, and its NVFP4 checkpoint runs on one NVIDIA B300 GPU</p> <p>The post <a href="https://www.marktechpost.com/2026/08/02/thinking-machines-lab-releases-inkling-small-276b-open-weights-multimodal-moe-model/">Thinking Machines La…
Thinking Machines Lab has released Inkling-Small, an open-weights multimodal MoE model with 276B total and 12B active parameters. The model beats its larger sibling on coding and reasoning benchmarks while running on a single NVIDIA B300 GPU. Available under Apache 2.0 on Hugging…
Thinking Machines releases Inkling Small, a smaller open-weights model that outperforms its predecessor on coding and reasoning tasks. Efficiency gains over size may signal a practical shift in model development. Source: The Decoder AI https:// the-decoder.com/thinking-machi nes-…