GLM 5.3 744B — B200 vs B300 Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus B300 (NVIDIA Blackwell) on GLM 5.3 744B. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. 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 →
Push GLM 5.3 744B to 169 tok/s/user and B200 lands at $0.08 per million tokens against B300's $0.08 — B200 pulls ahead by 2%.
B200: $0.10 per million tokens. B300: $0.12. Both at 213 tok/s/user on GLM 5.3 744B, with B200 17% cheaper.
Toward the upper edge of the 126–300 tok/s/user interactivity band — at 257 tok/s/user — B200 runs $0.13 per million tokens on GLM 5.3 744B while B300 runs $0.15. B200 is the cheaper choice by 20%. (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.)
Chip pricing (owning hyperscaler): B200 $1.73/chip/hr · B300 $2.26/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | B200:$0.082B300:$0.084 | B200:$0.101B300:$0.119 | B200:$0.125B300:$0.151 |
| Concurrency | B200:~7B300:~9 | B200:~5B300:~6 | B200:~4B300:~4 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.