Thinking Machines Unveils New AI Technology

2026-07-16 · Source: Artificial Intelligence: Educational AI News · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Emerging Technologies & Innovation · Depth: Intermediate, long

Summary

Thinking Machines has launched Inkling, a 975 billion-parameter open-weight AI model trained on 45 trillion tokens, achieving 84.7% on financial reasoning benchmarks at 1/14th the cost of top models. OpenAI introduced GPT-RED, an LLM super hacker that reduced successful attacks on GPT-5.6 from over 90% to 23%. AI Box released an MCP server integrating 80+ AI models into Claude, ChatGPT, and Gemini. AWS committed \$1 billion to custom AI engineering teams, mirroring OpenAI and Anthropic's forward-deployed strategies. Apple Intelligence gained China approval via a partnership with Alibaba's Quen. Meta plans to monetize its \$182.9 billion AI infrastructure by reselling compute capacity, adopting a SpaceX-like strategy.

Key takeaway

For AI Product Managers evaluating model deployment, consider open-weight models like Inkling for significant cost savings and customization, especially for domain-specific tasks. Your teams should also explore AI-powered red teaming for robust security and investigate integrating diverse AI models via platforms like AI Box MCP to enhance existing assistant capabilities. Additionally, assess opportunities to monetize excess compute infrastructure.

Key insights

AI innovation focuses on specialized models, security, integration, and strategic compute infrastructure.

Principles

• Specialized AI models often outperform general-purpose ones.
• AI-powered red teaming significantly enhances model security.
• Excess AI compute infrastructure can be monetized as a service.

Method

OpenAI's GPT-RED automates security flaw discovery by pitting an attacker model against a defender in simulated environments to harden systems.

In practice

• Download and fine-tune open-weight models like Inkling for custom applications.
• Integrate 80+ AI models into existing assistants via AI Box MCP.
• Explore reselling excess compute capacity for new revenue streams.

Topics

• AI Models
• Open-weight AI
• AI Security
• AI Compute Strategy
• Cloud AI Services
• Model Integration
• Fine-tuning

Best for: AI Engineer, Machine Learning Engineer, NLP Engineer, Director of AI/ML, AI Product Manager, Investor

Related on AIssential

• Thinking Machines Lab offers enterprises a US alternative in open-weight AI — Inkling offers Western enterprises a US-developed, open-weight, multimodal AI model for domain-specific adaptation.
• OpenAI is now using AI to attack its own AI, and it's working better than humans ever did — OpenAI's GPT-Red, an AI red teamer, significantly outperforms humans in finding prompt injection vulnerabilities through self-play reinforcement learning.
• Thinking Machines Lab’s First Model Gives Nod to Chinese Open Source AI — Thinking Machines Lab's Inkling model demonstrates leveraging established open-source architectures, specifically DeepSeek V3, for new AI development.
• INSANE AI News: GPT-RED, Kimi K3, Gemini 3.5 Pro and Anthropic's "END GAME" — Recursive Self-Improvement (RSI) and AI-driven safety mechanisms are rapidly advancing, shifting AI development paradigms.
• The Sequence Radar #897: Last Week in AI: China, Compression and the Open-Model Race — AI's future hinges on distributed access and governance, fragmenting the intelligence frontier beyond raw capability.
• not much happened today — The AI industry is rapidly integrating hardware, software, and agentic systems, intensifying competition and raising new infrastructure and security challenges.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence: Educational AI News.