gpt-oss 120B — Jalapeño vs MI300X
Head-to-head AI inference benchmark comparison of Jalapeño (OpenAI Jalapeño) and MI300X (AMD CDNA 3) 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.
Jalapeño / MI300X on gpt-oss 120B at 134 tok/s/user: 57705 / 3573 tok/s/chip, $0.01 / $0.07 per million tokens. Jalapeño is 1108% cheaper per token; Jalapeño delivers 1515% more tok/s/chip.
Around the middle of the 98–246 tok/s/user interactivity band, at 172 tok/s/user on gpt-oss 120B: Jalapeño runs 55160 tok/s/chip at $0.01/M tokens, MI300X runs 1871 at $0.14/M. Jalapeño is 2106% cheaper per token; Jalapeño delivers 2849% more tok/s/chip.
Setting 209 tok/s/user as the target on gpt-oss 120B, Jalapeño produces 52292 tok/s/chip ($0.01 per million tokens) and MI300X produces 898 ($0.29). Jalapeño is 4257% cheaper per token; Jalapeño delivers 5724% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | Jalapeño:57705.5MI300X:3572.6 | Jalapeño:55160.2MI300X:1870.6 | Jalapeño:52291.8MI300X:897.9 |
| Cost ($/M tok) | Jalapeño:$0.006MI300X:$0.074 | Jalapeño:$0.006MI300X:$0.141 | Jalapeño:$0.007MI300X:$0.294 |
| tok/s/MW | Jalapeño:51293776MI300X:2570199 | Jalapeño:49031314MI300X:1345770 | Jalapeño:46481589MI300X:645949 |
| Concurrency | Jalapeño:~1366MI300X:~6 | Jalapeño:~863MI300X:~7 | Jalapeño:~622MI300X:~4 |