LN-Tools — Ollama (v0.12.11+, MIT) is PUMA’s single local inference engine:…

Ollama (v0.12.11+, MIT) is PUMA’s single local inference engine: OpenAI-compatible REST, deterministic (seed/temperature), GGUF support, logprobs exposure (v0.12.11+) enabling calibration metrics; run as container puma_ollama on port 11434.

Source: PUMA project documentation · Traceability: corpus unit ANXAG-014 · Confidence: verified-at-primary-source