Open Models have crossed a threshold
Open models like GLM-5 and MiniMax M2.7 now match closed frontier models on core agent tasks — file operations, tool use, and instruction following — at a fraction of the cost and latency. Here's what our evals show and how to start using them in Deep Agents.
Open models like GLM-5 and MiniMax M2.7 now match closed frontier models on core agent tasks such as file operations, tool use, and instruction following, at a fraction of the cost and latency. This article explores the findings from evaluations and how to integrate these models into Deep Agents.
What happened
The LangChain Blog reports that open models such as GLM-5 and MiniMax M2.7 have reached parity with closed frontier models in core agent tasks like file operations, tool use, and instruction following. These models are more cost-effective and faster, with evaluations showing they can handle real-world workflows reliably.
Why it matters
For AI developers and businesses, open models provide a cost-effective and efficient alternative to expensive closed models. This shift can lead to more scalable and accessible AI solutions, especially for applications requiring real-time performance and lower operational costs.
- Open models like GLM-5 and MiniMax M2.7 now match closed models in core agent tasks.
- They offer significant cost and latency advantages, making them suitable for real-time applications.
- Deep Agents SDK simplifies the integration of open models into existing workflows.
- Open models can be used alongside or in place of closed models, depending on the use case.
- Evaluations show that open models are reliable and consistent for real-world workflows.
The bigger picture
The use of open models can significantly reduce costs and improve latency for AI applications. This makes them ideal for latency-sensitive products and high-throughput workloads, enabling more scalable and accessible AI solutions.
Open models are now a viable alternative to closed frontier models, offering cost and latency advantages without compromising performance. This shift can lead to more scalable and accessible AI solutions, especially for applications requiring real-time performance.
Technical details
- Evaluations were conducted using Deep Agents harness, covering tasks like file operations, tool use, and instruction following.
- Metrics such as correctness, solve rate, step ratio, and tool call ratio were used to assess model performance.
- Open models like GLM-5 and MiniMax M2.7 are available through providers like Baseten, Fireworks, Groq, OpenRouter, and Ollama.
- The Deep Agents SDK allows for easy integration of open models with minimal code changes.
FAQ
What are the key advantages of using open models over closed models?
Open models like GLM-5 and MiniMax M2.7 offer significant cost and latency advantages over closed models, making them suitable for real-time applications and high-throughput workloads.
How can I integrate open models into Deep Agents?
Integration is straightforward with the Deep Agents SDK, requiring only a one-line change to specify the model.
What metrics were used to evaluate the performance of open models?
Metrics such as correctness, solve rate, step ratio, and tool call ratio were used to assess model performance.
Which providers support open models for Deep Agents?
Providers like Baseten, Fireworks, Groq, OpenRouter, and Ollama support open models for Deep Agents.
Can open models be used alongside closed models?
Yes, open models can be used alongside or in place of closed models, depending on the specific use case and requirements.
This summary and analysis were generated with AI from the original article and may contain errors. Read the original for the full story.
Read the full story
Originally published by LangChain Blog on Oct 13, 2026.
Comments
Sign in to join the discussion.