Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and B300 (NVIDIA Blackwell) 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.
Near the low end of the 9–173 tok/s/user interactivity band, at 50 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B200 runs 8178 tok/s/chip at $0.06/M tokens, B300 runs 3504 at $0.18/M. B200 is 205% cheaper per token; B200 delivers 133% more tok/s/chip.
Setting 91 tok/s/user as the target on Kimi K2.5/K2.6/K2.7-Code 1T, B200 produces 1746 tok/s/chip ($0.28 per million tokens) and B300 produces 1963 ($0.32). B200 is 16% cheaper per token; B300 delivers 12% more tok/s/chip.
At 132 tok/s/user interactivity on Kimi K2.5/K2.6/K2.7-Code 1T, B200 delivers 962 tok/s/chip at $0.50 per million tokens; B300 delivers 857 tok/s/chip at $0.73. B200 is 47% cheaper per token; B200 delivers 12% more tok/s/chip at this point. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:8178.4B300:3504.2 | B200:1745.8B300:1962.6 | B200:961.9B300:856.6 |
| Cost ($/M tok) | B200:$0.059B300:$0.179 | B200:$0.275B300:$0.320 | B200:$0.500B300:$0.733 |
| tok/s/MW | B200:4782716B300:1844337 | B200:1020964B300:1032938 | B200:562522B300:450830 |
| Concurrency | B200:~323B300:~32 | B200:~62B300:~10 | B200:~30B300:~3 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr:
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
Kimi K2.5/2.6/2.7-Code 1T 8K / 1K Total Tokens per $1 TCO vs. Interactivity