Some ideas for what comes next, May 2026

2023-11-24 · Source: Interconnects AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Intermediate, medium

Summary

The AI landscape in May 2026 is characterized by accelerating capabilities, economic shifts, and escalating risks, with no slowdown expected. A significant gap persists between open and closed models; open-weight models have not achieved an "agent moment" like Opus 4.5 in Claude Code (December 2025), a performance level anticipated to take 12+ months for open alternatives. This suggests open models will likely specialize in automated enterprise agents, as even Google's Gemini lacks a direct competitor to advanced agentic models. An open-weights "Mythos" model is not expected this year, primarily due to resource constraints in Chinese labs versus the substantial compute available to major US companies. Conversely, American open models are gaining traction, with Gemma 4 outperforming Qwen 3.5/3.6 and permissive licenses boosting adoption. Anthropic and OpenAI are locked in fierce competition, rapidly advancing models like GPT 5.5 and Claude, which are transforming work. Simultaneously, existing power structures are increasingly asserting control over AI, leading to potential social and political conflicts.

Key takeaway

For AI Directors and ML Scientists navigating the evolving AI landscape, recognize that frontier closed models will continue to lead in agentic capabilities. Your open model strategies should prioritize specialization for automated enterprise agents and low-cost domains. Adopt permissive licenses for open-source initiatives to foster broader adoption. Prepare for increasing assertions of control from existing power structures and potential social conflicts, which may impact continued AI development. Engage in building a diverse open model ecosystem now to shape its future.

Key insights

Frontier closed models are outpacing open models, leading to specialization and escalating societal friction around AI.

Principles

• Real-world agentic utility is the true test for open model performance.
• Compute availability dictates frontier AI capability and progress.
• Permissive licenses drive open model adoption and brand building.

In practice

• Specialize open model development for automated enterprise agents.
• Utilize AI agents to refine and sell niche expertise.
• Adopt permissive licenses for open model releases.

Topics

• Open Models
• AI Agents
• Frontier AI
• AI Economics
• AI Governance
• Compute Resources

Best for: CTO, VP of Engineering/Data, AI Architect, AI Scientist, Director of AI/ML, Consultant

Related on AIssential

• What comes next with open models — Open models, while lagging closed counterparts, offer influence and cost-efficiency, especially as specialized tools complementing frontier AI.
• NVIDIA GTC 2026 Open Models Panel Highlights with Jensen Huang — Open models are becoming essential components in agentic systems, even for closed-model companies.
• Open models: Hot or Not with Nathan Lambert & Florian Brand — The open model ecosystem is saturated yet dynamic, with specialist models and licensing strategies significantly impacting adoption and competition.
• OpenAI, Anthropic fight on the frontier — Leading AI labs are rapidly advancing models with self-improvement, multi-agent capabilities, and deep enterprise integration.
• NVIDIA GTC 2026 Open Models Panel Highlights with Jensen Huang — AI's future lies in a "proprietary and open" ecosystem, leveraging compound agents and orchestrated multi-model systems.
• OpenAI launches a way for enterprises to build and manage AI agents — OpenAI Frontier provides an open, end-to-end platform for enterprise AI agent management, integrating external data and applications.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Interconnects AI.