🚀 FastFlowLM v0.9.38 - New Models!

FastFlowLM v0.9.38 adds qwen3.5:0.8b and nanbeige4.1:3b.

📦 New Models

🖼️ Qwen3.5:0.8B

FastFlowLM now supports a lightweight vision-language model from the Qwen3.5 family:

• Tag: qwen3.5:0.8b

You can control image pre-resizing in both run and serve modes with --img-pre-resize (or -r).

Run in CLI mode:

flm run qwen3.5:0.8b -r 1

Run in server mode:

flm serve qwen3.5:0.8b -r 1

For more details, see the model card and benchmark results.

---

🧠 Nanbeige4.1-3B

FastFlowLM now supports the new Nanbeige4.1-3B model, a reasoning model with tool-calling capabilities.

• Tag: nanbeige4.1:3b

Run in CLI mode:

flm run nanbeige4.1:3b

Run in server mode:

flm serve nanbeige4.1:3b

For more details, see the model card and benchmark results.

---

🐛 Bug Fixes

Some feature labels were missing from the model list JSON file, which could cause issues when integrating with downstream apps that rely on this list.

This issue is fixed in this release.

🌟 Summary

FastFlowLM v0.9.38 expands the Qwen3.5 lineup with Qwen3.5-0.8B and adds Nanbeige4.1-3B for reasoning and tool-calling workflows. Give both models a try and explore the latest performance improvements! ✨