Saturday, Aug 1, 2026
Added Quantized as a built-in provider to luca-py
Anurag Verma integrated Quantized as an OpenAI-compatible provider and made progress on native subagent rendering in the TUI.
1 person shipped this
Verma registered Quantized (api.quantized.us/v1) as a built-in provider in quantized-ai/luca-py [1], following the same one-entry configuration pattern already used by groq, deepseek, and ollama. The integration added 9 flagship model catalog entries including Claude 4.x, GPT-5.1, GPT-4o-mini, DeepSeek v3.2 and R1, Qwen3, and Llama 3.3, each with real context windows sourced from the Quantized API [2]. Quantized now appears in the /model TUI picker alongside the other providers, with documentation and tests included [2]. In parallel work, Verma began refactoring subagent rendering to key the mount target by conversation_id rather than positional attribution, so that each subagent gets its own titled SubagentPanel [3]. He also scripted a single subagent into the --faux offline demo, focusing on the multiply gates pattern to exercise the approval modal's conversation-context display [4].
Sources
- Add quantized as a built-in provider · quantized-ai/luca-py
- Add quantized as a built-in OpenAI-compatible provider (#15) · quantized-ai/luca-py
- Render subagents natively in the TUI · quantized-ai/luca-py
- Script a subagent into the --faux demo · quantized-ai/luca-py