🤖AI Agents Weekly: LLMs in 2025, YOLO in the Sandbox, Plan Caching for Agents, DeepTutor

2026-01-03 · Source: AI Newsletter · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Software Development & Engineering · Depth: Advanced, quick

Summary

Simon Willison's annual review, "2025: The Year in LLMs," details 26 significant trends that shaped the Large Language Model landscape in 2025. Key developments include the breakthrough of coding agents, with Claude Code achieving $1 billion in run-rate revenue by December, and other major labs releasing their own CLI coding agents like Codex CLI and Gemini CLI. Reasoning models proved crucial for driving multi-step agents, enabling effective AI-assisted search and complex code debugging. Chinese models, such as GLM-4.7 and DeepSeek V3.2, dominated open-weight rankings, with DeepSeek R1's January release notably impacting NVIDIA's market cap by $593 billion. The report also highlights the normalization of risky AI practices, termed "YOLO mode," and the rapid increase in long task capabilities, with AI task duration doubling every 7 months.

Key takeaway

For Machine Learning Engineers evaluating LLM trends, recognize that 2025 saw coding agents become a major revenue driver and reasoning models unlock advanced agent capabilities. You should prioritize integrating robust reasoning models into your agent designs to handle complex, multi-step tasks and consider the implications of "YOLO mode" for deployment safety, especially as AI task duration continues to rapidly expand.

Key insights

Coding agents and reasoning models significantly advanced LLM capabilities and market impact in 2025.

Principles

• Reasoning models drive effective tool use.
• Open-weight models can disrupt market leaders.

In practice

• Explore CLI coding agents for development tasks.
• Monitor Chinese open-weight models for performance.

Topics

• Large Language Models
• AI Agents
• Coding Agents
• Reasoning Models
• AI Safety

Best for: Machine Learning Engineer, NLP Engineer, CTO, AI Engineer, AI Researcher, AI Product Manager

Related on AIssential

• The last six months in LLMs in five minutes — LLM capabilities, particularly coding agents and local models, advanced rapidly from November 2025 to April 2026, marked by frequent "best" model shifts.
• Fundamental knowledge SWE's in 2026 must have before I will even consider hiring them — Building an LLM agent involves a simple loop integrating specialized tools and careful context management.
• Top 10 LLM Research Papers of 2026 — LLM research in 2026 prioritizes safety, control, and agentic utility over mere scale.
• What’s Next in AI: Five Trends to Watch in 2026 — AI's 2026 trajectory is defined by the convergence of advanced reasoning, production-ready agents, specialized coding, open-weight models, and multimodal capabilities.
• AI Trends 2026: OpenClaw Agents, Reasoning LLMs, and More with Sebastian Raschka - #762 — LLM progress is shifting from raw model scaling to reasoning-focused post-training, inference-time techniques, and better tool integration.
• Are Tools All We Need? Unveiling the Tool-Use Tax in LLM Agents — Tool-use in LLM agents incurs a "tool-use tax" that can negate performance gains, especially with semantic noise.

Open in AIssential →

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