📊 Full opportunity report: Decoding AI's Future: What Thinking Machines’ Inkling Reveals on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Thinking Machines launched Inkling, a 975-billion-parameter open-access AI model, openly acknowledging it is not the most powerful available. The release emphasizes transparency and ownership, but raises questions about licensing and use policies.
Thinking Machines has publicly released its first foundation model, Inkling, a 975-billion-parameter multimodal transformer, openly available on Hugging Face under the Apache 2.0 license. This marks a notable shift in AI deployment, emphasizing transparency and ownership over proprietary secrecy.
The Inkling model is a mixture-of-experts transformer supporting a 1-million-token context window, trained on 45 trillion tokens across text, images, audio, and video. It was developed by Thinking Machines Lab, founded by former OpenAI CTO, and staffed with engineers involved in ChatGPT’s development.
The model is available with full weights on Hugging Face under an Apache 2.0 license, allowing users to download, modify, and deploy independently. Unlike typical commercial models, Inkling’s release prioritizes openness, but the company has reportedly implemented a separate use policy restricting surveillance, deception, and automated decision-making affecting individuals’ rights. The licensing and policy details are not fully disclosed, prompting questions about scope and enforceability.
Despite its openness, Inkling is not the most powerful model currently available. It ranks mid-tier on several benchmarks but excels in safety and multimodal capabilities, such as speech recognition and adversarial robustness. The model’s training involved hybrid optimization techniques and reinforcement learning with synthetic data generated by other open-weight models.
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.
Impact of Open-Access AI Model Release
This release represents a significant shift toward transparency and ownership in AI development, challenging the dominance of proprietary models. It allows organizations to independently fine-tune, inspect, and deploy the model, potentially accelerating innovation and reducing dependency on closed platforms. However, the layered use restrictions and lack of full training data transparency raise questions about true openness and the potential for misuse. The development underscores ongoing tensions between open-source ideals and responsible AI deployment, impacting how future models may be governed and adopted.
AI Engineering: Building Applications with Foundation Models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background on AI Model Releases and Industry Norms
In recent months, major AI labs have largely released models with limited access or closed licenses, citing safety concerns and commercial interests. The release of Inkling under Apache 2.0, with full weights, marks a departure from this trend, emphasizing open access. Historically, open models have faced challenges related to misuse, safety, and intellectual property, leading many to restrict access or impose layered policies. Thinking Machines’ approach reflects a growing debate over balancing openness with responsible use, especially as models grow larger and more capable.
The company’s transparency about training methods and benchmark results is relatively rare, and its candid acknowledgment that Inkling is not the strongest model available signals a focus on openness over competitive dominance. This approach may influence future industry standards for model releases, emphasizing transparency and user control.
“Our goal is to foster an ecosystem where AI can be owned, inspected, and responsibly deployed by anyone.”
— Thinking Machines spokesperson
Multimodal AI Systems Engineering: Building Production Vision-Language Models, Document AI, and Cross-Modal Retrieval Pipelines (Production AI Engineering Series)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Uncertainties Surrounding Inkling’s Use Policies
Details about Thinking Machines’ Model Acceptable Use Policy remain unclear. Reports suggest restrictions on surveillance, deception, and automated decision-making, but the full scope and enforceability of these restrictions are not confirmed. The impact of these policies on the model’s open license and practical use is still uncertain, raising questions for potential users about legal and ethical boundaries.
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.
Next Steps for Adoption and Evaluation
Independent researchers and organizations will likely conduct further benchmarking and real-world testing of Inkling’s capabilities and safety features. Attention will focus on how the layered use policies are enforced and how the model performs across diverse applications. Additionally, the release may prompt other labs to reconsider their licensing strategies, possibly leading to more open models or layered policies. Monitoring how the ecosystem responds will be critical in understanding Inkling’s impact on AI development.
AI Engineering and Agentic AI: Designing Autonomous Language Model Systems with Memory, Tools, and Safe Deployment
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What makes Inkling different from other AI models?
Inkling is openly available under the Apache 2.0 license, supports multimodal input, and emphasizes transparency and ownership, unlike many proprietary models.
Is Inkling the most powerful model available?
No, Inkling ranks mid-tier on several benchmarks but is notable for its safety features and multimodal capabilities.
What restrictions are placed on Inkling’s use?
Reports suggest there are restrictions on surveillance, deception, and automated decisions affecting individuals, but full details are not confirmed.
Why does open licensing matter in AI?
Open licensing allows users to freely download, modify, and deploy models, fostering innovation and reducing reliance on closed platforms.
What are the risks of open models like Inkling?
Open models can be misused for malicious purposes, and layered policies may complicate enforcement and ethical oversight.
Source: ThorstenMeyerAI.com
