Codaipro Files

CodaiPro v2.1.1

Maintenance release: everything that broke while the project slept has been repaired. Same product, working installs, smarter AI output.

Fixed

• Fresh installs work again: fastapi, uvicorn, requests and pydantic are now declared (they were imported but never listed)
• AI engine upgraded to llama-cpp-python 0.3.x; removed the deprecated f16_kv and FastAPI on_event usage
• Every model now receives its own chat template — Phi-3.5 and Qwen2.5-Coder answers are no longer degraded by a hardcoded Phi-3 prompt
• Single-instance guard now watches the real backend port (8765, was 8000) and lockfiles are cleaned on exit
• The GUI no longer spawns a second backend alongside the launcher's
• Generation timeout raised to 300 s so slow lab machines stop showing fake errors
• GPU option actually works: CUDA build flag fixed, CODAIPRO_GPU_LAYERS=99 offloads layers
• Website redesigned and live at https://luckyyaduvanshiofficial.github.io/Codai/ — with a real poll and live star count

Download options

• Portable Windows: unzip, run CodaiPro_v21.exe, no installation — perfect for lab environments and USB sticks
• Source code: clone and pip install -r requirements.txt

First run

1. Download a model (Phi-3.5-mini 2.3 GB, Qwen2.5-Coder-3B 2 GB, or Qwen2.5-Coder-7B 4.7 GB) into the models folder — links in the model guide
2. Run CodaiPro_v21.exe

System requirements

• Windows 10/11 (64-bit)
• 4 GB RAM minimum
• 2 GB free storage + 2–5 GB for one model

Full changelog: https://github.com/Luckyyaduvanshiofficial/Codai/compare/git-tag-v2.1.0...v2.1.1