DeepSeek-V4-Flash-0731 Q8_K_XL sometimes stops mid-task in OpenCode - anyone else seeing this?

Article automatically generated from technical news.

Hey everyone, I've been experimenting with the new DeepSeek-V4-Flash-0731 release locally using the Unsloth Studio Q8_K_XL GGUF with OpenCode. Overall, it's been working really well, but I've noticed a strange behavior during longer agentic coding sessions. Once the context gets above ~100K tokens, the model will sometimes be in the middle of thinking/working through a task and then just stop generating. There doesn't seem to be an obvious error or crash. It just stops. If I ty