● FitLLM — fit receipt · Q4_K_M · 8K tokens ctx · KV F16 · engine 2.8.1

Qwen 3.8 27B on RTX 3090

FITS ✓

weights	KV cache	linear state	overhead	reserve	total	free
15.8 GB	0.5 GB	0.1 GB	2.2 GB	2.0 GB	20.7 / 24 GB	3.3 GB

Max context at this quant: ~29K tokens

Replay: npx fitllm-engine "Qwen 3.8 27B" --gpu "RTX 3090" · JSON · interactive

live fit badge — embed: ![fits](https://img.shields.io/endpoint?url=https%3A%2F%2Ffitllm.run%2Fapi%2Fbadge%3Fmodel%3DQwen%2B3.8%2B27B%26gpu%3DRTX%2B3090%26quant%3DQ4_K_M%26ctx%3D8192%26kv%3D16)

Computed by the open fitllm-engine (MIT) from official config.json values — estimates; runtime varies. Ran it for real? Challenge this prediction — measured reports calibrate the engine for everyone.