Open-weight 952B-parameter language model from Thinking Machines Lab, its first public open-weight release.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Inkling is a large language model from Thinking Machines Lab, the AI research and product company. According to a creator review, it is the lab's first public open-weight model and has 952 billion parameters. The lab's site states it plans to share code, datasets and model specs, but the provided official pages do not include a model card for Inkling.
Reviewers describe its benchmark results as decent but behind GLM 5.2 and leading closed models. One creator tried it for free in the Tinker playground, the lab's tooling for customizing and fine-tuning models. Details such as context length, architecture, modalities and license are not confirmed in the provided material.
Access is through Thinking Machines' platform, and weights are reported to be downloadable. Because of its size, self-hosting requires substantial multi-GPU infrastructure.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
First public open-weight model from Thinking Machines, 952B parameters. Benchmarks are decent but behind GLM 5.2 and closed models; the creator tried it free in the Tinker playground.
Closest neighbours by meaning, not just by category.
Model · Thinking Machines
Open-weights multimodal MoE model (276B total, 12B active) from Thinking Machines; text, image and audio in, text out.
Model · Thinking Machines
Open-weights multimodal MoE model (594B total, 17B active) from Thinking Machines, available via Tinker.
Model · Alibaba
Alibaba's Qwen 3.5 large open-weights language model (~122B parameters) from the Qwen family.
Model · OpenAI
OpenAI's open-weight reasoning models, gpt-oss-120b and gpt-oss-20b, released under Apache 2.0.
Model
Compact 1.7B open-weights text LLM for on-device use, with adjustable thinking levels and English/Chinese support.
Model
Open 7.39B dense model for math reasoning and agentic search, with a 256K context and the full training pipeline released.
