Deploy Open Models on GPUs

Choose a GPU setup to find the open models that fit, and deploy one as an API.

1× NVIDIA RTX PRO 6000 Blackwell

Verified models first, then by average benchmark score.

#	Model	Params	Precision	Max context
1	Qwen3.8-27B 2026.08	27.8B	FP8	256K
2	Qwen-AgentWorld-35B-A3B 2026.06	34.7B	FP16	256K
3	Gemma 4 31B 2026.03	31B	FP16	256K
4	GLM-4.7-Flash 2026.01	31B	FP16	198K
5	UI-Mate-27B 2026.08	27.4B	FP16	256K
6	TaskMem 2026.06	31.1B	FP16	240K
7	ZAYA1-8B 2026.05	8.8B	FP8	128K
8	Kimi-VL-A3B-Thinking-2506 2025.06	16.4B	FP16	128K
9	Cosmos3-Edge 2026.07	3.9B	FP16	128K
10	Fara-7B 2025.11	8.3B	FP16	125K
11	GLM-4.1V-9B-Thinking 2025.07	10.3B	FP16	64K
12	Nemotron 3 Nano Omni 2026.04	33B	NVFP4	256K
13	Mage-VL 2026.07	4.7B	FP16	256K
14	North-Micro-Vision-Instruct 2026.08	2.5B	FP16	488K