DeepSeek Releases Its Own Harness and V4 Pro
DeepSeek has introduced a new harness and the V4 Pro model, notable for its modularity and quick component swapping capabilities.
DeepSeek Releases Its Own Harness and V4 Pro
The main feature of the DeepSeek Harness is extreme modularity. Almost everything is a module with hot reload capability, allowing for dynamic changes to model connectors and the agent loop itself. A detailed paper has been released explaining how this works.
Alongside this, they have released the V4 Pro from preview, traditionally releasing weights under the MIT license.
As promised, the company is raising prices starting August 16. Input tokens will now be 1.5x more expensive, while output tokens will be more than 2x. Additionally, during peak hours, prices will double again.
Why it matters
AnalysisThis is significant as the new technologies from DeepSeek could greatly enhance the modularity and flexibility of AI model usage. The price increase also indicates a growing demand for their services.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Six DeepSeek Harness Plugins to Cut Token Usage and Organize Context
A review of six plugins for DeepSeek Harness, covering free model access, memory organization, agent routing, and context monitoring.
DeepSeek launches Creator Mode to let users reshape their harness
DeepSeek’s open-source harness now supports Creator Mode, where an agent can write and embed plugins based on plain-language descriptions.
DeepSeek (Chinese Breakthrough in Open AI)
A series of groundbreaking open models from the Chinese lab DeepSeek (V3, R1). Demonstrates top-tier coding and reasoning capabilities at a cost 10–20 times lower than Western commercial counterparts.
How to Run Local LLMs: Setup Guide, Model Benchmarks, and Optimization
The definitive guide to running local LLMs on your hardware: hardware requirements, Ollama and Open WebUI deployment, Llama 3, Qwen 2.5, DeepSeek, and GGUF quantization.