Kimi K3 2.8T · Chip comparison

Kimi K3 2.8T — B300 vs H200

Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) 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

Throughput at 52 tok/s/user on Kimi K3 2.8T: B300 hits 5944 tok/s/chip, H200 hits 70. Per-million costs land at $0.11 and $4.85 respectively. B300 is 4493% cheaper per token; B300 delivers 8408% more tok/s/chip.

B300 / H200 on Kimi K3 2.8T at 102 tok/s/user: 3999 / 70 tok/s/chip, $0.16 / $4.85 per million tokens. B300 is 2989% cheaper per token; B300 delivers 5623% more tok/s/chip.

Toward the upper edge of the 2–202 tok/s/user interactivity band, at 152 tok/s/user on Kimi K3 2.8T: B300 runs 2072 tok/s/chip at $0.30/M tokens, H200 runs 70 at $4.85/M. B300 is 1501% cheaper per token; B300 delivers 2865% 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)
B300:5944.5H200:69.9
B300:3998.8H200:69.9
B300:2071.8H200:69.9
Cost ($/M tok)
B300:$0.106H200:$4.850
B300:$0.157H200:$4.850
B300:$0.303H200:$4.850
tok/s/MW
B300:3128669H200:51001
B300:2104612H200:51001
B300:1090443H200:51001
Concurrency
B300:~10H200:~1
B300:~6H200:~1
B300:~2H200:~1

Inference Performance

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