Models
Lucid Train is local-first and model-agnostic. Run open-source models on your own hardware, bring an OpenRouter key for the frontier, or point it at a self-hosted endpoint.
Local · Ollama (free, offline)
/download <tag> installs Ollama automatically if needed, pulls the model, and tunes it for low memory pressure (flash attention, q8_0 KV cache, models unload between turns). The setup screen marks each model ✓ fits, ◐ low free RAM, or ✗ too big for your machine.
| Model | Needs | Notes |
|---|---|---|
| gpt-oss:20b | ~16 GB RAM | Best tool-calling that fits 16 GB · the default recommendation. |
| qwen3-coder:30b | ~32 GB RAM | Best local agentic coder (RL-trained on SWE-Bench). |
| qwen3:4b | ~6 GB RAM | Great small all-rounder for modest machines. |
| deepseek-r1 / ERNIE 4.5 | varies | Strong reasoning specialists (chat-first). |
Cloud · OpenRouter (bring your own key)
One key unlocks the whole catalog. Your key is stored locally and only ever sent to OpenRouter. ★ Lucid's choice is kimi-k2.7-code · a coding specialist that balances quality and price. On a budget, deepseek-v4-pro is the cheapest near-frontier option.
| Model | Context | $/1M in→out | Notes |
|---|---|---|---|
| kimi-k2.7-code | 262K | $0.75 → $3.50 | ★ Lucid's choice · coding specialist |
| deepseek-v4-pro | 1M | $0.44 → $0.87 | cheapest near-frontier, 80.6% SWE-bench |
| qwen3.7-max | 1M | $1.25 → $3.75 | long-horizon agentic |
| kimi-k2.7 / k2.6 | 262K | ~$0.75 → $3.50 | general Kimi line |
| glm / minimax | 200K | varies | GLM 5.1 · MiniMax M2.7 |
| gpt-5.1 / grok / gemini | 200K-1M | premium | proprietary frontier |
Self-hosted (OpenAI-compatible)
Point Lucid Train at any vLLM, llama.cpp, SGLang, or TGI server:
Failover
List backup models and the client walks them on provider errors: