ArticleThinking Machines LabAnnouncement
Inkling-Small demonstrates how post-training distillation and specialized agentic reinforcement learning can make smaller open-weights models outperform much larger systems in software engineering and reasoning at significantly lower inference costs.
By deploying on-policy distillation from the 975B parameter Inkling and extending agentic coding RL, Thinking Machines Lab built a highly efficient generalist model. Benchmarks such as Terminal-Bench 2.1 and SWEBench-Verified highlight strong execution on agentic workflows relative to its compute footprint. However, the drop in factuality on benchmarks like SimpleQA confirms that parameter reduction primarily impacts memorized knowledge rather than logical and tool-based problem-solving capabilities.
Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.
Similar in content, found by the AI search.
Article:Thinking Machines Lab
Thinking Machines Lab has introduced Inkling, an open-weights Mixture-of-Experts model featuring 975B total parameters (41B active) and a 1M token context window. It is built as a generalist foundation optimized for customization via the Tinker platform.
Link:inclusionAI
inclusionAI has introduced Ling-3.0-tiny, a compact MoE model with 7.9B total parameters activating only 1.3B parameters per token. Utilizing a hybrid architecture of linear and latent attention mechanisms, it aims to deliver efficient reasoning and agentic workloads on local hardware.
Repository:inclusionAI/Ling
InclusionAI has open-sourced Ling, a family of Mixture-of-Experts (MoE) large language models. The suite is offered in two sizes: Ling-lite with 16.8 billion total parameters (2.75 billion activated) and Ling-plus with 290 billion total parameters (28.8 billion activated).
267starsPython
Link:ornith-ai
ornith-ai has released Ornith-1.5-397B, an open-source multimodal mixture-of-experts model under the MIT licence. With 397 billion parameters and an end-to-end self-improvement loop, it delivers coding and reasoning benchmark performance comparable to top proprietary models.
Article:Zhidx
Ant Group's Bailian team has released Ling-3.0-tiny, an open-source lightweight mixture-of-experts model. It leverages DeepSeek's MLA and Kimi's KDA architectures with 7.9B total parameters, activating just 1.3B parameters per token for local edge deployment on hardware like Apple Silicon Macs and DGX systems.
Article:Chris Alexiuk and Chintan Patel
NVIDIA has announced Nemotron 3.5 Lightning, an open 30B mixture-of-experts (MoE) model featuring 3B active parameters. It is tailored for the high-volume execution layer of long-running AI agents, targeting routine tasks such as tool calling and result validation with low latency.
Parts of this content were created with the assistance of AI.
