📊 Full opportunity report: The Significance Of Initial Inkling From Thinking Machines In AI Progress on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Thinking Machines has publicly released its first foundation model, Inkling, with open weights under Apache 2.0. This move emphasizes transparency and ownership in AI, sparking industry discussions about open-source practices and model governance.
Thinking Machines has officially released its first foundation model, Inkling, under open licensing, marking a significant step in transparency and ownership in AI development. This is the first time the company has made the full model weights publicly available, allowing for independent inspection, modification, and deployment, which is a notable departure from typical industry practices.
The Inkling model is a 975-billion-parameter Mixture-of-Experts transformer supporting a 1-million-token context window. It was trained on 45 trillion tokens across text, images, audio, and video, and features a native multimodal input design. The model’s weights are available on Hugging Face under the Apache 2.0 license, enabling broad use, modification, and commercial deployment.
Unlike many large models, Thinking Machines did not restrict access via a closed API but instead released the full weights upfront. This approach emphasizes transparency and user ownership, addressing concerns raised after recent government-mandated shutdowns of frontier models. However, the company also reportedly maintains a separate Model Acceptable Use Policy restricting surveillance, deception, and automated decision-making, which complicates the open-source narrative.
The weights came first: what Inkling actually signals
Mira Murati’s lab shipped its first foundation model — and the model isn’t the story. The order of operations is: full weights, Apache 2.0, day one, before any closed API. Plus a rare concession — the lab says it’s not the strongest model available, open or closed.
- AIME 2026 97.1%
- GPQA Diamond 87.2%
- MCP Atlas (Nemotron 44.7%) 74.1%
- VoiceBench · open-weight audio frontier 91.4%
- FORTRESS adversarial · best open 78.0%
- ForecastBench · calibration 61.1
- HLE text-only (GLM-5.2 40.1%) 29.7%
- SWE-bench Pro (GLM-5.2 62.1%) 54.3%
- Terminal-Bench 2.1 (GLM-5.2 82.7%) 63.8%
- SWE-bench Verified (Fable 5 95.0%) 77.6%
- Design Arena · 2nd open, behind GLM-5.2 ~10th
A 0.2 → 0.99 effort setting trades reasoning tokens against cost & latency, so you get a curve, not a point. On Terminal-Bench 2.1 it reportedly matches Nemotron 3 Ultra at ~⅓ the tokens. Peak score is a vanity metric when you serve millions of calls; the cost curve is what ships. (Bonus: its chain of thought compressed on its own during RL — nobody rewarded it; efficiency did.)
Pitched as the Western alternative to Chinese open weights (censorship-resistance training is the differentiator). But GLM-5.2 still wins on agentic/reasoning and Kimi K2.6 often on multimodal: best American open model, second in the open field. The irony — post-training was bootstrapped on synthetic data from Kimi K2.5.
BF16 needs ≥2 TB aggregate VRAM (8× B300 / 16× H200). NVFP4 still needs ≥600 GB. Not a workstation model — a 512 GB fleet falls just short. “Open” ≠ “runnable.” Mitigations: 1-bit GGUFs (~74% acc.), hosted eval routes, and Inkling-Small (12B active) — the release local-first builders actually want.
Open weights used to be a consolation prize. Inkling is a strategic open release — Apache 2.0, natively multimodal, honestly marketed, published complete on day one, optimized for deployment rather than headlines (the model isn’t the product; the fine-tuning platform is). It doesn’t need to win every benchmark for that to matter. The frontier is learning that owning the base beats renting the API — arriving now from the inside. For the sovereignty buyer: ① a real Western hedge against being switched off · ② verify the use policy before you build · ③ check the VRAM, then benchmark vs GLM-5.2 & Kimi K2.6 on your task.
Implications of Open-Weight Release in AI Industry
The release of Inkling’s full weights under open license represents a shift toward greater transparency and ownership in AI development. It allows organizations to independently inspect, modify, and deploy the model, reducing reliance on third-party API providers. This move could influence industry standards, encouraging more open practices and potentially leading to a more decentralized AI ecosystem. However, the presence of a separate Use Policy raises questions about the true openness and enforceability of restrictions on the model’s use, which could impact trust and adoption.
AI Systems Performance Engineering: Optimizing Model Training and Inference Workloads with GPUs, CUDA, and PyTorch
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background on Open Models and Industry Shift
Historically, most large AI models have been released with restricted access, often via APIs, to control usage and manage safety concerns. Recent industry and regulatory developments have increased scrutiny on AI transparency and ownership, especially after incidents where models were shut down abruptly due to policy or government action. The April 2024 release of Inkling by Thinking Machines, a company staffed by former OpenAI engineers, signals a potential shift towards more open, owner-controlled models. This aligns with broader debates about open-source AI, safety, and the balance between transparency and misuse prevention.
“Our goal is to empower users with ownership and transparency, while maintaining responsible use through our policies.”
— Thinking Machines spokesperson
multimodal AI development tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unclear Aspects of Open Model Governance
It is still unclear how effective the separate Model Acceptable Use Policy will be in practice, and whether it will significantly restrict the model’s open use. The scope and enforceability of these restrictions, especially in comparison to the freedoms granted by Apache 2.0, remain uncertain. Additionally, the full training data and pipeline have not been published, raising questions about transparency in the training process.
large language model datasets
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for Industry Adoption and Scrutiny
Industry observers will likely monitor how organizations and developers utilize Inkling in real-world applications, especially regarding compliance with the Use Policy. Independent benchmarks and replication efforts will assess the model’s performance and safety. The company may also clarify or update its policies based on community feedback and regulatory developments. Further releases, including the upcoming smaller version, Inkling-Small, are expected to demonstrate the model’s capabilities and limitations.
AI model deployment software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What makes Inkling different from other large language models?
Inkling is notable for its full open weights under Apache 2.0 license, allowing unrestricted download, modification, and deployment, unlike most models that are API-restricted or closed-source.
Does open access mean the model is completely free of restrictions?
No. While the weights are open, Thinking Machines reportedly maintains a separate Use Policy that restricts certain applications, such as surveillance and deceptive practices. The enforceability of these restrictions remains to be seen.
Why is the release of Inkling significant for AI development?
It signals a shift toward greater transparency and ownership in AI, potentially encouraging more open practices and reducing reliance on API-based models, which could reshape industry standards.
What are the potential risks of open-sourcing such a large model?
Open-sourcing large models raises concerns about misuse, safety, and compliance, especially if restrictions are layered separately and not fully transparent. It also complicates efforts to enforce responsible use.
What is the next step for the company regarding Inkling?
The company will likely release testing benchmarks, clarify policy details, and support community evaluation to build trust and understand the model’s capabilities and limitations better.
Source: ThorstenMeyerAI.com
