gpt-oss 120B — H200 vs Jalapeño
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and Jalapeño (OpenAI Jalapeño) on gpt-oss 120B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
Throughput at 137 tok/s/user on gpt-oss 120B: H200 hits 4834 tok/s/chip, Jalapeño hits 57522. Per-million costs land at $0.07 and $0.01 respectively. Jalapeño is 1043% cheaper per token; Jalapeño delivers 1090% more tok/s/chip.
H200 / Jalapeño on gpt-oss 120B at 176 tok/s/user: 3604 / 54874 tok/s/chip, $0.09 / $0.01 per million tokens. Jalapeño is 1362% cheaper per token; Jalapeño delivers 1422% more tok/s/chip.
Toward the upper edge of the 98–255 tok/s/user interactivity band, at 216 tok/s/user on gpt-oss 120B: H200 runs 2306 tok/s/chip at $0.15/M tokens, Jalapeño runs 51665 at $0.01/M. Jalapeño is 2052% cheaper per token; Jalapeño delivers 2140% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H200:4834.4Jalapeño:57521.6 | H200:3604.4Jalapeño:54874.5 | H200:2306.3Jalapeño:51664.8 |
| Cost ($/M tok) | H200:$0.070Jalapeño:$0.006 | H200:$0.094Jalapeño:$0.006 | H200:$0.147Jalapeño:$0.007 |
| tok/s/MW | H200:3528768Jalapeño:51130334 | H200:2630977Jalapeño:48777326 | H200:1683406Jalapeño:45924229 |
| Concurrency | H200:~16Jalapeño:~1316 | H200:~10Jalapeño:~828 | H200:~5Jalapeño:~588 |