MiniMax M3 428B · Chip comparison

MiniMax M3 428B — B300 vs H100

Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H100 (NVIDIA Hopper) on MiniMax M3 428B. 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 hits 42,346 tok/s/chip for $0.015 per million tokens at 152 tok/s/user on MiniMax M3 428B. No H100 data at this operating point.

B300: 15,288 tok/s/chip, $0.041 per million tokens at 248 tok/s/user on MiniMax M3 428B. H100 is unmeasured here.

At 343 tok/s/user on MiniMax M3 428B, B300 delivers 5,054 tok/s/chip at $0.12 per million tokens; H100 hasn't been benchmarked at this target. (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 →

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)
B30042,346H100—
B30015,288H100—
B3005,054H100—
Cost ($/M tok)
B300$0.015H100—
B300$0.041H100—
B300$0.12H100—
tok/s/MW
B30022,287,395H100—
B3008,046,521H100—
B3002,659,949H100—
Concurrency
B300~25H100—
B300~9H100—
B300~5H100—