Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Kimi K2.5/K2.6/K2.7-Code 1T. 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.
B300 / MI355X on Kimi K2.5/K2.6/K2.7-Code 1T at 38 tok/s/user: 4317 / 4004 tok/s/chip, $0.15 / $0.10 per million tokens. MI355X is 40% cheaper per token; B300 delivers 8% more tok/s/chip.
Around the middle of the 11–123 tok/s/user interactivity band, at 67 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B300 runs 2747 tok/s/chip at $0.23/M tokens, MI355X runs 2493 at $0.17/M. MI355X is 37% cheaper per token; B300 delivers 10% more tok/s/chip.
Setting 95 tok/s/user as the target on Kimi K2.5/K2.6/K2.7-Code 1T, B300 produces 1848 tok/s/chip ($0.34 per million tokens) and MI355X produces 1739 ($0.24). MI355X is 42% cheaper per token; B300 delivers 6% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:4316.9MI355X:4003.6 | B300:2746.6MI355X:2492.7 | B300:1847.7MI355X:1738.7 |
| Cost ($/M tok) | B300:$0.145MI355X:$0.104 | B300:$0.229MI355X:$0.167 | B300:$0.340MI355X:$0.240 |
| tok/s/MW | B300:2272062MI355X:1915595 | B300:1445598MI355X:1192676 | B300:972471MI355X:831906 |
| Concurrency | B300:~51MI355X:~102 | B300:~19MI355X:~17 | B300:~9MI355X:~9 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr: VR200: 3.61H100: 1.17H200: 1.22B200: 1.73B300: 2.26GB200: 1.86GB300: 2.31MI300X: 0.95MI325X: 1.1MI355X: 1.5RTX6000PRO: 0.68Jalapeño (Teacup): 1.47
Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate cost per million tokens per decode chip or per prefill chip, rather than per total chip count. This makes direct token cost comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate input throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct input throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate output throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct output throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate power per decode chip or per prefill chip, rather than per total chip count. This makes direct power comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate Joules per decode chip or per prefill chip, rather than per total chip count. This makes direct Joules per token comparison with aggregated configs not an apples-to-apples comparison.
Matching measurements exist, but their chip series are hidden.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip
1 The ATOM engine is promising, however it has yet to serve production tokens. It is still in its infant stage.
Kimi K2.5/2.6/2.7-Code 1T 8K / 1K Total Tokens per $1 TCO vs. Interactivity