RTX Pro 6000 96GB

// weights AWQ 4-bit for Mixtral 8x22B (141B params)
weights = params × bits ÷ 8
        = 141 × 4.25 ÷ 8
        = 74.91 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
74.91 ≤ 79.05  → FITS
headroom  = 4.14 GB of weights budget left

// weights Q4_K_M for gpt-oss 120B (117B params)
weights = params × bits ÷ 8
        = 117 × 4.5 ÷ 8
        = 65.81 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
65.81 ≤ 79.05  → FITS
headroom  = 13.24 GB of weights budget left

// weights Q5_K_M for Command R+ (104B params)
weights = params × bits ÷ 8
        = 104 × 5.5 ÷ 8
        = 71.50 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
71.50 ≤ 79.05  → FITS
headroom  = 7.55 GB of weights budget left

// weights FP8/INT8 for Qwen 2.5 72B (72.7B params)
weights = params × bits ÷ 8
        = 72.7 × 8 ÷ 8
        = 72.70 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
72.70 ≤ 79.05  → FITS
headroom  = 6.35 GB of weights budget left

// weights Q8_0 for Llama 3.3 70B (70.6B params)
weights = params × bits ÷ 8
        = 70.6 × 8.5 ÷ 8
        = 75.01 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
75.01 ≤ 79.05  → FITS
headroom  = 4.04 GB of weights budget left

// weights Q8_0 for Mixtral 8x7B (46.7B params)
weights = params × bits ÷ 8
        = 46.7 × 8.5 ÷ 8
        = 49.62 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
49.62 ≤ 79.05  → FITS
headroom  = 29.43 GB of weights budget left

// weights FP16/BF16 for Qwen 3.6 35B A3B (35B params)
weights = params × bits ÷ 8
        = 35 × 16 ÷ 8
        = 70.00 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
70.00 ≤ 79.05  → FITS
headroom  = 9.05 GB of weights budget left

// weights FP16/BF16 for Yi 34B (34B params)
weights = params × bits ÷ 8
        = 34 × 16 ÷ 8
        = 68.00 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
68.00 ≤ 79.05  → FITS
headroom  = 11.05 GB of weights budget left

// weights FP16/BF16 for Qwen 2.5 32B (32.5B params)
weights = params × bits ÷ 8
        = 32.5 × 16 ÷ 8
        = 65.00 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
65.00 ≤ 79.05  → FITS
headroom  = 14.05 GB of weights budget left

// weights FP16/BF16 for Qwen 2.5 Coder 32B (32.5B params)
weights = params × bits ÷ 8
        = 32.5 × 16 ÷ 8
        = 65.00 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
65.00 ≤ 79.05  → FITS
headroom  = 14.05 GB of weights budget left

// weights FP16/BF16 for Qwen3 30B A3B (30.5B params)
weights = params × bits ÷ 8
        = 30.5 × 16 ÷ 8
        = 61.00 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
61.00 ≤ 79.05  → FITS
headroom  = 18.05 GB of weights budget left

// weights FP16/BF16 for Qwen 3.6 27B (27B params)
weights = params × bits ÷ 8
        = 27 × 16 ÷ 8
        = 54.00 GB

// budget on RTX Pro 6000 (96GB) at ctx 8K, conc 1, 15% safety
kv_cache  = 0.05 GB    (1× at ctx 8K)
overhead  = 2.50 GB    (runtime, cuda, allocator)
safety    = 14.40 GB    (15% of 96GB)
budget    = vram − safety − kv − overhead
          = 96 − 14.40 − 0.05 − 2.50
          = 79.05 GB

// fit decision
54.00 ≤ 79.05  → FITS
headroom  = 25.05 GB of weights budget left