RTX Pro 6000 Blackwell

58 records across 7 models (quants: MXFP4, Q2 LynnStyle, Q3 LynnStyle, Q4 LynnStyle, Q4_0, Q4_K, Q4_K - Medium, Q6_K, Q8_0).

hardware	model	quant	backend	tok/s	pp tok/s	W	ttft s	ctx/tg	date	class	flags	source
RTX Pro 6000 Blackwell	Gemma4 26B 26B	Q4_K	llama.cpp	196.9	9437.2	–	–	4096	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 26B 26B	Q4_K	llama.cpp	172.7	8453.8	–	–	16384	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 26B 26B	Q4_K	llama.cpp	170.3	7107.6	–	–	32768	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 26B 26B	Q4_K	llama.cpp	161.0	5379.6	–	–	65536	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 26B 26B	Q4_K	llama.cpp	133.2	3667.8	–	–	131072	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 26B 26B	Q4_K	llama.cpp	112.5	2245.7	–	–	262144	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 31B 31B	Q4_K	llama.cpp	61.6	3749.8	–	–	4096	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 31B 31B	Q4_K	llama.cpp	59.6	3061.9	–	–	16384	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 31B 31B	Q4_K	llama.cpp	55.8	2086.7	–	–	32768	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 31B 31B	Q4_K	llama.cpp	52.0	1423.0	–	–	65536	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 31B 31B	Q4_K	llama.cpp	43.6	876.8	–	–	131072	–	sourced	–	source
RTX Pro 6000 Blackwell	Gemma4 31B 31B	Q4_K	llama.cpp	34.4	506.8	–	–	262144	–	sourced	–	source
RTX Pro 6000 Blackwell	gpt-oss 120B 120B	MXFP4	llama.cpp	210.1	4578.6	–	–	4096	–	sourced	–	source
RTX Pro 6000 Blackwell	gpt-oss 120B 120B	MXFP4	llama.cpp	182.2	4060.7	–	–	16384	–	sourced	–	source
RTX Pro 6000 Blackwell	gpt-oss 120B 120B	MXFP4	llama.cpp	161.5	3368.3	–	–	32768	–	sourced	–	source
RTX Pro 6000 Blackwell	gpt-oss 120B 120B	MXFP4	llama.cpp	133.1	2360.3	–	–	65536	–	sourced	–	source
RTX Pro 6000 Blackwell	gpt-oss 120B 120B	MXFP4	llama.cpp	99.8	1289.8	–	–	131072	–	sourced	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	289.3	14463.16	–	–	tg128	2026-06-11	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	287.29	13403.7	–	–	tg128	2026-05-08	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	281.11	16618.98	–	–	tg128	2025-09-05	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	275.79	18012.64	–	–	tg128	2026-04-02	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	275.76	16652.18	–	–	tg128	2025-08-11	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	274.2	14854.63	–	–	tg128	2025-09-05	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	271.26	12242.46	–	–	tg128	2026-05-08	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	268.52	14401.14	–	–	tg128	2025-08-11	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	264.66	15928.07	–	–	tg128	2026-04-02	community	–	source
RTX Pro 6000 Blackwell	llama 7B	Q4_0	llama.cpp (CUDA)	260.42	12742.48	–	–	tg128	2026-06-11	community	–	source
RTX Pro 6000 Blackwell	nemotron_h_moe 31B.A3.5B	Q4_K - Medium	llama.cpp	447.63	–	–	–	8192	2026-10-01	community	–	source
RTX Pro 6000 Blackwell	nemotron_h_moe 31B.A3.5B	Q4_K - Medium	llama.cpp	432.58	–	–	–	8192	2026-10-01	community	–	source
RTX Pro 6000 Blackwell	nemotron_h_moe 31B.A3.5B	Q4_K - Medium	llama.cpp	314.14	–	–	–	–	2026-10-01	community	–	source
RTX Pro 6000 Blackwell	nemotron_h_moe 31B.A3.5B	Q4_K - Medium	llama.cpp	313.28	–	–	–	–	2026-10-01	community	–	source
RTX Pro 6000 Blackwell	nemotron_h_moe 31B.A3.5B	Q4_K - Medium	llama.cpp	–	11485.61	–	–	–	2026-10-01	community	–	source
RTX Pro 6000 Blackwell	nemotron_h_moe 31B.A3.5B	Q4_K - Medium	llama.cpp	–	11474.42	–	–	–	2026-10-01	community	–	source
RTX Pro 6000 Blackwell	Qwen3.5-27B 27B	Q4_K	llama.cpp	60.9	3338.5	–	–	4096	–	sourced	–	source
RTX Pro 6000 Blackwell	Qwen3.5-27B 27B	Q4_K	llama.cpp	57.5	2980.5	–	–	16384	–	sourced	–	source
RTX Pro 6000 Blackwell	Qwen3.5-27B 27B	Q4_K	llama.cpp	55.1	2526.7	–	–	32768	–	sourced	–	source
RTX Pro 6000 Blackwell	Qwen3.5-27B 27B	Q4_K	llama.cpp	51.3	1856.7	–	–	65536	–	sourced	–	source
RTX Pro 6000 Blackwell	Qwen3.5-27B 27B	Q4_K	llama.cpp	45.1	1404.9	–	–	131072	–	sourced	–	source
RTX Pro 6000 Blackwell	Qwen3.5-27B 27B	Q4_K	llama.cpp	36.3	903.1	–	–	262144	–	sourced	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q2 LynnStyle	NInfer	521.0	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q3 LynnStyle	NInfer	512.3	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q8_0	NInfer	486.3	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q4 LynnStyle	NInfer	436.4	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	NInfer	388.1	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	NInfer	235.5	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q3 LynnStyle	llama.cpp (CUDA)	191.8	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q8_0	llama.cpp (CUDA)	190.8	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	llama.cpp (CUDA)	189.0	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q2 LynnStyle	llama.cpp (CUDA)	174.8	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	NInfer	131.9	–	–	–	32768	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	llama.cpp (CUDA)	119.0	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q3 LynnStyle	llama.cpp (CUDA)	112.2	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q3 LynnStyle	llama.cpp (CUDA)	110.2	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	llama.cpp (CUDA)	97.2	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q3 LynnStyle	llama.cpp (CUDA)	95.6	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q3 LynnStyle	llama.cpp (CUDA)	94.4	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q4 LynnStyle	llama.cpp (CUDA)	81.1	–	–	–	–	2026-10-08	community	–	source
RTX Pro 6000 Blackwell	Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO	Q6_K	llama.cpp (CUDA)	53.4	–	–	–	–	2026-10-08	community	–	source

All rows cite their source. Provenance: community 40018 sourced 1085

← all hardware