Просмотр новости
Найдите то, что Вас интересует
Спутники
РН
SpaceX
Роскосмос
NASA
Thinking Machines Lab Releases Inkling, a 975 Billion Parameter Open Weights AI Model Under Apache 2.0
Дата публикации: 16-07-2026 08:50:06
Thinking Machines Lab released Inkling on Wednesday, an AI model with 975 billion parameters.
Thank you for being a Ghacks reader. The post Thinking Machines Lab Releases Inkling, a 975 Billion Parameter Open Weights AI Model Under Apache 2.0 appeared first on gHacks.
Основное содержимое страницы с новостью.
Thinking Machines Lab released Inkling on Wednesday, an AI model with 975 billion parameters. The company, founded in early 2025 by former OpenAI CTO Mira Murati, is providing the model under the Apache 2.0 license, which allows developers to freely fine-tune and use it.
Inkling is currently the largest open-weights model from the United States and is positioned as an alternative to Chinese open-weights models such as DeepSeek V4, GLM 5.2, and Kimi K2.6. It is available now on Thinking Machines' Tinker platform and via Hugging Face for direct download.
Model Specs, Hardware Requirements, and Positioning Against Chinese Open WeightsInkling combines a mixture of expert architecture with the following features:
- 975 billion total parameters
- 256 routed experts, and two shared experts
- Six experts activate for each token, resulting in around 41 billion parameters during inference. It supports a token context window of one million
- The model was trained on 45 trillion tokens encompassing text, images, audio, and video and was developed using Nvidia GB300 NVL72 systems
While inspired by DeepSeek-V3, Inkling was trained from scratch by Thinking Machines, rather than fine-tuned from existing weights.
Running Inkling at native 16-bit precision requires more than two terabytes of GPU memory. Practical hardware setups include about eight Nvidia B300 accelerators or sixteen Nvidia H200 accelerators.
For users with less GPU capacity, Thinking Machines offers an NVFP4 quantized version of the model that requires approximately half the GPUs. This version trades some precision for reduced memory usage while maintaining most of the model's capabilities.
Reasoning Capabilities, Apache 2.0 Licensing, and Developer AccessLike most modern frontier models, Inkling is a reasoning model trained with reinforcement learning to use chain-of-thought thinking before responding.
Thinking Machines claims Inkling matches Nvidia's Nemotron 3 Ultra, which was previously the largest American open weights model at 550 billion parameters.
It reportedly achieves comparable results on Terminal Bench 2.1 using roughly one-third the thinking tokens. This efficiency means Inkling could deliver similar performance at a lower cost when priced by tokens.
Users should note that thinking tokens are billed like any other tokens, so longer chains of thought can increase the cost of each response. Inkling enters a market where Chinese labs have dominated open-weight AI development. Competitors include:
- DeepSeek V4 and V4-Pro from Huawei Ascend,
- GLM 5.2, and
- Kimi K2.6.
Thinking Machines claims Inkling is competitive with these Chinese models across various workloads. However, users are advised to verify such claims independently through their own testing, as gaming industry benchmarks for AI are often unreliable.
The benchmark charts for Inkling also show it trailing behind proprietary models such as Anthropic's Claude and OpenAI's GPT. The key tradeoff for users lies in gaining access to weights and full customization versus relying on closed models with higher raw performance.
Fine-tuning, self-modification, and access points for Inkling are also part of the broader discussion.
Thinking Machines describes Inkling as highly adaptable for developers building AI applications and general-purpose uses like chatbots. The Apache 2.0 license allows commercial use, modification, and redistribution.
The company also claims that Inkling can write its own fine-tuning scripts to refine its behavior, teach itself new skills, and evaluate its abilities. The self-modification capability is aimed at making customization more accessible to developers without deep machine learning expertise.
The company's Tinker platform provides tools for customization and fine-tuning. For developers wanting to use Inkling:
- Tinker platform: Available now for API access and fine-tuning tools
- Hugging Face: Model weights available for direct download
- Third-party API providers: Thinking Machines is working with TogetherAI, Fireworks, Modal, Databricks, and Baseten to add the model to their services
Inkling supports a broad range of inference engines at launch including vLLM, SGLang, Miles, TokenSpeed, and Llama.cpp.
What Users Should Do, Inkling-Small Preview, and AvailabilityFor developers interested in Inkling for their projects: Review the model's benchmark performance in relation to your specific use case rather than relying solely on general claims. Consider whether the 1 million token context window justifies the hardware requirements for your workload.
If full precision hardware isn't feasible, evaluate the quantized NVFP4 version. Test the model through the Tinker platform before committing to a self-hosted deployment. It's also useful to compare token efficiency claims with real-world workloads in your domain.
For users exploring Inkling for hobby or evaluation purposes: Download the model via Hugging Face. Set up an inference engine such as Llama.cpp for local testing. Be prepared for significant hardware requirements, even with quantization.
Alongside Inkling, Thinking Machines is previewing Inkling-Small, a 276 billion-parameter mixture of experts (MoE) model with 12 billion active parameters.
This smaller model targets users who prioritize low latency over throughput and quality. It has not yet been released, but the company plans to release the weights once testing concludes. A specific timeline for release has not been provided.
Inkling’s launch comes at a time when the frontier AI market is increasingly dominated by proprietary models with limited access. OpenAI recently postponed the public release of GPT-5.6 at the request of the US government.
Anthropic's Fable 5 and Mythos 5 were suspended temporarily before being restored. Alibaba also restricted access to Claude Code for employees over tracking concerns.
In this context, Thinking Machines’ decision to release Inkling under the Apache 2.0 license offers an alternative for developers who want access to frontier-scale capabilities without relying on a single provider. The name references the fictional supercomputer maker from Jurassic Park.
Inkling is now available on the Tinker platform and Hugging Face. Third-party API access through platforms like TogetherAI, Fireworks, Modal, Databricks, and Baseten is expected to follow.
The company has not yet announced pricing for the Tinker platform. Interested users can follow the company's official channels for updates on Inkling-Small and future releases.
Схожие новости
Классификация: Пресс-релизы. Схожих патентов: 0. Схожих новостей: 9. Тональность: 0. Информативность: 7.35. Источник: www.ghacks.net.