Spark-X2.6-4B Released with Stronger Coding and Agent Skills

Key Info

SparkLLM announced Spark-X2.6-4B, a new 4B-parameter model whose post-training improvements target coding, agentic tasks, and long-context skills.

Highlights

• Model size: 4B parameters
• Improved post-training for coding, agents, and long-context performance
• Thinking modes: none / low / high
• Shared with the Hugging Face, ModelScope, Ollama, vLLM, and SGLang communities