What comes next with open models

2023-11-24 · Source: Interconnects AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, AI Business Strategy · Depth: Advanced, long

Summary

In 2025, many companies began adopting open models to gain influence in the AI ecosystem, a trend accelerated by the success of models like DeepSeek R1. Despite the high cost of participation, open models offer a fast path to mindshare and usage without extensive marketing. However, open models consistently lag behind closed models by 6-18 months, a gap that is likely to widen as frontier AI development shifts to specialized, non-public data domains and complex RL environments for agents. The article projects a future with three classes of models: true (closed) frontier models for advanced knowledge work, open frontier models for general use cases where absolute best performance isn't critical, and open, small, specialized models designed for niche, repetitive tasks. The most successful open models will act as complementary tools to closed agents, offering cost and speed advantages for specific functions.

Key takeaway

For CTOs and AI Engineers evaluating their model strategy, recognize that while closed models will lead in frontier performance, open models offer significant value in specialized, cost-efficient applications. Prioritize developing or integrating small, task-specific open models as complementary tools to larger AI systems, rather than attempting to directly compete with closed models on general benchmarks. This approach can drive innovation and reduce operational costs for repetitive tasks.

Key insights

Open models, while lagging closed counterparts, offer influence and cost-efficiency, especially as specialized tools complementing frontier AI.

Principles

• Open models trail closed models by 6-18 months.
• AI systems integrate weights, tools, and harnesses.
• Open systems foster innovation and competition.

Method

Developing open models should focus on creating specialized, cheap, and fast tools for niche tasks, rather than directly competing with closed frontier models on general capabilities, to complement larger AI systems.

In practice

• Build small, specialized models with LoRA adapters.
• Outsource repetitive agent tasks to small open models.
• Focus on compute/time savings as benchmarks.

Topics

• Open Models
• Closed Models
• AI Ecosystem
• Model Performance Gap
• Specialized AI Models

Best for: CTO, VP of Engineering/Data, AI Engineer, Director of AI/ML, AI Product Manager, AI Architect

Related on AIssential

• Some ideas for what comes next, May 2026 — Frontier closed models are outpacing open models, leading to specialization and escalating societal friction around AI.
• The Thriving Ecosystem of Open Models — The open-model ecosystem is rapidly expanding, driven by competition and new releases, capturing significant developer token volume on platforms like OpenRouter.
• 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.
• TAI #208: Open Models Find Their Role as Agent Token Bills Rise — Tiered AI model deployment optimizes agent costs by assigning tasks to models based on capability and price.
• NVIDIA GTC 2026 Open Models Panel Highlights with Jensen Huang — Open models are becoming essential components in agentic systems, even for closed-model companies.
• 🎙️Nathan Lambert: Open Models Will Never Catch Up — Open models, though trailing frontier systems, are vital for AI research and innovation, especially in academia.

Open in AIssential →

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