Kimi K3 2.8T · Chip comparison

Kimi K3 2.8T — B300 vs GB300 NVL72

Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on Kimi K3 2.8T. 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.

AgentX replays real coding-agent sessions rather than fixed-length prompts, so context grows turn over turn and most of each request is served from cache instead of being recomputed. That turns the comparison into a systems question: KV transfer between nodes, prefix-aware routing, and cache capacity all move the curve alongside raw chip throughput. Fixed-sequence workloads stay the clean baseline for kernel and silicon performance, so the two scenarios answer different questions about the same hardware. Learn more about AgentX

B300 / GB300 NVL72 on Kimi K3 2.8T at 57 tok/s/user: 5615 / 7833 tok/s/chip, $0.11 / $0.08 per million tokens. GB300 NVL72 is 36% cheaper per token; GB300 NVL72 delivers 39% more tok/s/chip.

Around the middle of the 8–202 tok/s/user interactivity band, at 105 tok/s/user on Kimi K3 2.8T: B300 runs 3911 tok/s/chip at $0.16/M tokens, GB300 NVL72 runs 5133 at $0.13/M. GB300 NVL72 is 28% cheaper per token; GB300 NVL72 delivers 31% more tok/s/chip.

Setting 154 tok/s/user as the target on Kimi K3 2.8T, B300 produces 2069 tok/s/chip ($0.30 per million tokens) and GB300 NVL72 produces 3391 ($0.19). GB300 NVL72 is 60% cheaper per token; GB300 NVL72 delivers 64% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)

View performance-per-dollar view →

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Throughput (tok/s/chip)
B300:5615.2GB300 NVL72:7833.0
B300:3911.3GB300 NVL72:5132.6
B300:2068.8GB300 NVL72:3390.6
Cost ($/M tok)
B300:$0.112GB300 NVL72:$0.082
B300:$0.161GB300 NVL72:$0.125
B300:$0.303GB300 NVL72:$0.189
tok/s/MW
B300:2955360GB300 NVL72:3694809
B300:2058569GB300 NVL72:2421025
B300:1088859GB300 NVL72:1599352
Concurrency
B300:~9GB300 NVL72:~63
B300:~5GB300 NVL72:~35
B300:~2GB300 NVL72:~8

Inference Performance

Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.