The B200 192GB SXM is a NVIDIA accelerator with 192 GB HBM3e and about 8000 GB/s peak memory bandwidth. AI Hashrate estimates local LLM decode (batch 1) at 8K context: fit counts weights + KV cache + 1 GB runtime overhead — not weights alone. At Q4, about 50 curated models fit fully on this card; at FP16, about 44 fit. Top Q4 speeds: gemma-3-1b-it ≈ 5000 tok/s (estimated, fits); Qwen3-0.6B ≈ 5000 tok/s (estimated, fits); Llama-3.2-1B-Instruct ≈ 4242.4 tok/s (estimated, fits); Qwen3-1.7B ≈ 2545.5 tok/s (estimated, fits); Qwen3-30B-A3B ≈ 1697.0 tok/s (estimated, fits). Relative ranking is more reliable than absolute tok/s. See methodology for the bandwidth formula and measured-anchor policy. Methodology.
| Model | Params (B) | Quant | tok/s | Fits? |
|---|---|---|---|---|
| gemma-3-1b-it | 1.0 | Q4 | 5000 est. | Yes |
| Qwen3-0.6B | 0.8 | Q4 | 5000 est. | Yes |
| Llama-3.2-1B-Instruct | 1.2 | Q4 | 4242.4 est. | Yes |
| Qwen3-1.7B | 2.0 | Q4 | 2545.5 est. | Yes |
| Qwen3-0.6B | 0.8 | FP16 | 1750.0 est. | Yes |
| Qwen3-30B-A3B | 30.5 | Q4 | 1697.0 est. | Yes |
| Qwen3-Coder-30B-A3B-Instruct | 30.5 | Q4 | 1697.0 est. | Yes |
| Qwen3-30B-A3B-Instruct-2507 | 30.5 | Q4 | 1697.0 est. | Yes |
| Qwen3-Next-80B-A3B-Instruct | 81.3 | Q4 | 1697.0 est. | Yes |
| Qwen3-Coder-Next | 79.7 | Q4 | 1697.0 est. | Yes |
| Llama-3.2-3B-Instruct | 3.2 | Q4 | 1590.9 est. | Yes |
| gpt-oss-20b | 21.5 | Q4 | 1414.1 est. | Yes |
| gemma-3-1b-it | 1.0 | FP16 | 1400.0 est. | Yes |
| Phi-4-mini-instruct | 3.8 | Q4 | 1339.7 est. | Yes |
| Phi-4-mini-reasoning | 3.8 | Q4 | 1339.7 est. | Yes |
| gemma-3-4b-it | 4.3 | Q4 | 1305.4 est. | Yes |
| Qwen3-4B-Instruct-2507 | 4.0 | Q4 | 1272.7 est. | Yes |
| Qwen3-4B-Thinking-2507 | 4.0 | Q4 | 1272.7 est. | Yes |
| Llama-3.2-1B-Instruct | 1.2 | FP16 | 1166.7 est. | Yes |
| Mistral-7B-Instruct-v0.3 | 7.2 | Q4 | 707.1 est. | Yes |
| Qwen3-1.7B | 2.0 | FP16 | 700.0 est. | Yes |
| Qwen2.5-7B-Instruct | 7.6 | Q4 | 669.9 est. | Yes |
| Qwen2.5-Coder-7B-Instruct | 7.6 | Q4 | 669.9 est. | Yes |
| DeepSeek-R1-Distill-Qwen-7B | 7.6 | Q4 | 669.9 est. | Yes |
| Llama-3.1-8B-Instruct | 8.0 | Q4 | 636.4 est. | Yes |
| gpt-oss-120b | 120.4 | Q4 | 636.4 est. | Yes |
| Qwen3-8B | 8.2 | Q4 | 620.8 est. | Yes |
| DeepSeek-R1-0528-Qwen3-8B | 8.2 | Q4 | 620.8 est. | Yes |
| Llama-3.1-70B-Instruct | 70.6 | Q4 | 520.0 measured (nvidia-gtc) | Yes |
| MiniMax-M2.7 | 228.7 | Q4 | 509.1 est. | Yes |
| Qwen3-30B-A3B | 30.5 | FP16 | 466.7 est. | Yes |
| Qwen3-Coder-30B-A3B-Instruct | 30.5 | FP16 | 466.7 est. | Yes |
| Qwen3-30B-A3B-Instruct-2507 | 30.5 | FP16 | 466.7 est. | Yes |
| Qwen3-Next-80B-A3B-Instruct | 81.3 | FP16 | 466.7 est. | Yes |
| Qwen3-Coder-Next | 79.7 | FP16 | 466.7 est. | Yes |
| Llama-3.2-3B-Instruct | 3.2 | FP16 | 437.5 est. | Yes |
| gemma-3-12b-it | 12.2 | Q4 | 424.2 est. | Yes |
| GLM-4.7-Flash | 31.2 | Q4 | 424.2 est. | Yes |
| GLM-4.5-Air | 110.5 | Q4 | 424.2 est. | Yes |
| GLM-4.5 | 358.3 | Q4 | 400.8 est. | No |
| Mixtral-8x7B-Instruct-v0.1 | 46.7 | Q4 | 394.6 est. | Yes |
| DeepSeek-V4-Flash | 158.1 | Q4 | 391.6 est. | Yes |
| gpt-oss-20b | 21.5 | FP16 | 388.9 est. | Yes |
| Phi-4-mini-instruct | 3.8 | FP16 | 368.4 est. | Yes |
| Phi-4-mini-reasoning | 3.8 | FP16 | 368.4 est. | Yes |
| gemma-3-4b-it | 4.3 | FP16 | 359.0 est. | Yes |
| Qwen3-4B-Instruct-2507 | 4.0 | FP16 | 350.0 est. | Yes |
| Qwen3-4B-Thinking-2507 | 4.0 | FP16 | 350.0 est. | Yes |
| Phi-4 | 14.7 | Q4 | 346.3 est. | Yes |
| Phi-4-reasoning | 14.7 | Q4 | 346.3 est. | Yes |
| Qwen3-14B | 14.8 | Q4 | 344.0 est. | Yes |
| Qwen2.5-14B-Instruct | 14.8 | Q4 | 344.0 est. | Yes |
| Qwen2.5-Coder-14B-Instruct | 14.8 | Q4 | 344.0 est. | Yes |
| DeepSeek-R1-Distill-Qwen-14B | 14.8 | Q4 | 344.0 est. | Yes |
| gpt-oss-120b | 120.4 | FP16 | 311.1 est. | No |
| GLM-4.5-Air | 110.5 | FP16 | 244.0 est. | No |
| Qwen3-235B-A22B | 235.1 | Q4 | 231.4 est. | Yes |
| Qwen3-235B-A22B-Instruct-2507 | 235.1 | Q4 | 231.4 est. | Yes |
| Devstral-Small-2507 | 23.6 | Q4 | 215.7 est. | Yes |
| Mistral-Small-3.2-24B-Instruct-2506 | 24.0 | Q4 | 212.1 est. | Yes |
| Magistral-Small-2509 | 24.0 | Q4 | 212.1 est. | Yes |
| Qwen3-Coder-480B-A35B-Instruct | 480.2 | Q4 | 206.9 est. | No |
| Mistral-7B-Instruct-v0.3 | 7.2 | FP16 | 194.4 est. | Yes |
| gemma-3-27b-it | 27.4 | Q4 | 187.2 est. | Yes |
| Qwen2.5-7B-Instruct | 7.6 | FP16 | 184.2 est. | Yes |
| Qwen2.5-Coder-7B-Instruct | 7.6 | FP16 | 184.2 est. | Yes |
| DeepSeek-R1-Distill-Qwen-7B | 7.6 | FP16 | 184.2 est. | Yes |
| Llama-3.1-8B-Instruct | 8.0 | FP16 | 175.0 est. | Yes |
| MiniMax-M1 | 456.1 | Q4 | 171.4 est. | No |
| Qwen3-8B | 8.2 | FP16 | 170.7 est. | Yes |
| DeepSeek-R1-0528-Qwen3-8B | 8.2 | FP16 | 170.7 est. | Yes |
| Qwen3-32B | 32.8 | Q4 | 155.2 est. | Yes |
| Qwen2.5-32B-Instruct | 32.8 | Q4 | 155.2 est. | Yes |
| Qwen2.5-Coder-32B-Instruct | 32.8 | Q4 | 155.2 est. | Yes |
| DeepSeek-R1-Distill-Qwen-32B | 32.8 | Q4 | 155.2 est. | Yes |
| gemma-3-12b-it | 12.2 | FP16 | 116.7 est. | Yes |
| GLM-4.7-Flash | 31.2 | FP16 | 116.7 est. | Yes |
| DeepSeek-V4-Flash | 158.1 | FP16 | 110.9 est. | No |
| Mixtral-8x7B-Instruct-v0.1 | 46.7 | FP16 | 108.5 est. | Yes |
| DeepSeek-R1 | 684.5 | Q4 | 97.8 est. | No |
| DeepSeek-V3.2 | 685.4 | Q4 | 97.6 est. | No |
| Phi-4 | 14.7 | FP16 | 95.2 est. | Yes |
| Phi-4-reasoning | 14.7 | FP16 | 95.2 est. | Yes |
| Qwen3-14B | 14.8 | FP16 | 94.6 est. | Yes |
| Qwen2.5-14B-Instruct | 14.8 | FP16 | 94.6 est. | Yes |
| Qwen2.5-Coder-14B-Instruct | 14.8 | FP16 | 94.6 est. | Yes |
| DeepSeek-R1-Distill-Qwen-14B | 14.8 | FP16 | 94.6 est. | Yes |
| GLM-5.2 | 753.3 | Q4 | 74.8 est. | No |
| GLM-5.1 | 753.9 | Q4 | 74.7 est. | No |
| Llama-3.3-70B-Instruct | 70.6 | Q4 | 72.7 est. | Yes |
| DeepSeek-R1-Distill-Llama-70B | 70.6 | Q4 | 72.1 est. | Yes |
| Qwen2.5-72B-Instruct | 72.7 | Q4 | 70.0 est. | Yes |
| MiniMax-M2.7 | 228.7 | FP16 | 69.6 est. | No |
| Devstral-Small-2507 | 23.6 | FP16 | 59.3 est. | Yes |
| Mistral-Small-3.2-24B-Instruct-2506 | 24.0 | FP16 | 58.3 est. | Yes |
| Magistral-Small-2509 | 24.0 | FP16 | 58.3 est. | Yes |
| gemma-3-27b-it | 27.4 | FP16 | 51.5 est. | Yes |
| Kimi-K2-Instruct-0905 | 1026.5 | Q4 | 51.2 est. | No |
| Kimi-K2-Thinking | 1058.1 | Q4 | 48.2 est. | No |
| DeepSeek-V4-Pro | 861.6 | Q4 | 46.6 est. | No |
| Qwen3-32B | 32.8 | FP16 | 42.7 est. | Yes |
| Qwen2.5-32B-Instruct | 32.8 | FP16 | 42.7 est. | Yes |
| Qwen2.5-Coder-32B-Instruct | 32.8 | FP16 | 42.7 est. | Yes |
| DeepSeek-R1-Distill-Qwen-32B | 32.8 | FP16 | 42.7 est. | Yes |
| Qwen3-235B-A22B | 235.1 | FP16 | 29.6 est. | No |
| Qwen3-235B-A22B-Instruct-2507 | 235.1 | FP16 | 29.6 est. | No |
| Llama-3.1-70B-Instruct | 70.6 | FP16 | 20.0 est. | Yes |
| Llama-3.3-70B-Instruct | 70.6 | FP16 | 20.0 est. | Yes |
| DeepSeek-R1-Distill-Llama-70B | 70.6 | FP16 | 19.8 est. | Yes |
| Qwen2.5-72B-Instruct | 72.7 | FP16 | 19.3 est. | Yes |
| Llama-3.1-405B-Instruct | 405.9 | Q4 | 14.2 est. | No |
| GLM-4.5 | 358.3 | FP16 | 8.8 est. | No |