← AI Hashrate

B200 192GB SXM vs Gaudi 3 128GB for Local LLMs

The B200 192GB SXM has about 144% more memory bandwidth (8,000 vs 3,276 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 144% faster on the B200 192GB SXM. The B200 192GB SXM fits more of the catalog at Q4/8K — 50 vs 47 of 61 models (192 vs 128 GB VRAM). Largest model that fits the B200 192GB SXM but not the Gaudi 3 128GB: Qwen3-235B-A22B (235.1B). MSRP is $40,000 for the B200 192GB SXM vs $12,500 for the Gaudi 3 128GB. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. Not retail-shoppable: B200 192GB SXM, Gaudi 3 128GB (datacenter-class; cloud rental is the realistic way to use it). In the Q4/8K head-to-head below (59 shared models), the B200 192GB SXM posts the higher tok/s in 56 rows vs 3. Methodology.

Spec comparison

B200 192GB SXMGaudi 3 128GB
VendorNVIDIAIntel
VRAM192 GB128 GB
Memory typeHBM3eHBM2e
Memory bandwidth8,000 GB/s3,276 GB/s
TDP1000 W600 W
MSRP (list)$40,000$12,500
Buy——

MSRP is the launch list price, not live retail. Buy links are Amazon affiliate links — the price after you click is the live price.

Q4 / 8K head-to-head — top 15 of 59 shared models

ModelParams (B)B200 192GB SXM tok/sB200 192GB SXM fitsGaudi 3 128GB tok/sGaudi 3 128GB fitsWinner
gemma-3-1b-it1.05000 est.Yes2084.7 est.YesB200 192GB SXM
Qwen3-0.6B0.85000 est.Yes2605.9 est.YesB200 192GB SXM
Llama-3.2-1B-Instruct1.24242.4 est.Yes1737.3 est.YesB200 192GB SXM
Qwen3-1.7B2.02545.5 est.Yes1042.4 est.YesB200 192GB SXM
Qwen3-30B-A3B30.51697.0 est.Yes694.9 est.YesB200 192GB SXM
Qwen3-Coder-30B-A3B-Instruct30.51697.0 est.Yes694.9 est.YesB200 192GB SXM
GLM-4.7-Flash31.21697.0 est.Yes694.9 est.YesB200 192GB SXM
Qwen3-30B-A3B-Instruct-250730.51697.0 est.Yes694.9 est.YesB200 192GB SXM
Qwen3-Next-80B-A3B-Instruct81.31697.0 est.Yes694.9 est.YesB200 192GB SXM
Qwen3-Coder-Next79.71697.0 est.Yes694.9 est.YesB200 192GB SXM
Llama-3.2-3B-Instruct3.21590.9 est.Yes651.5 est.YesB200 192GB SXM
gpt-oss-20b21.51414.1 est.Yes579.1 est.YesB200 192GB SXM
Phi-4-mini-instruct3.81339.7 est.Yes548.6 est.YesB200 192GB SXM
Phi-4-mini-reasoning3.81339.7 est.Yes548.6 est.YesB200 192GB SXM
gemma-3-4b-it4.31305.4 est.Yes534.5 est.YesB200 192GB SXM

est. = bandwidth-based estimate; measured = curated real benchmark that overrides the estimate. Fits = weights + KV(8K) + 1 GB overhead ≤ 95% of VRAM.

Fit coverage at Q4 / 8K

B200 192GB SXMGaudi 3 128GB
Catalog models that fit50 of 6147 of 61
Largest exclusive fitQwen3-235B-A22B (235.1B)—

B200 192GB SXM details → · Gaudi 3 128GB details →