All model and GPU pairings
gpt-oss 120BNVIDIA Blackwell

Running gpt-oss-120b on B200

Quick answer

gpt-oss-120b sustains 44,489 tokens/s per GPU on B200 at 50 tokens/s per user, which works out to $0.011 per million tokens at hyperscaler pricing, served by vLLM. Fastest measured TTFT: 0.0 ms; fastest TPOT: 0.0 ms (each the best across all configs, not one run).

Benchmarked configs

98

Serving engines

TRTLLM, vLLM

Precisions

fp4

Run dates

2026-02-052026-05-31

Throughput at every interactivity target

Serving is a trade-off: push more concurrent users through a GPU and each user's tokens arrive slower. The ladder below reads the measured frontier at each per-user speed target on a single-turn chat workload (8k input / 1k output), using the best engine and precision at that point.

Per-user targetTokens/s per GPU$ / 1M tokensEnginePrecision
30 tok/s--vLLMfp4
50 tok/s44,489$0.011vLLMfp4
75 tok/s34,672$0.014vLLMfp4
100 tok/s29,548$0.016vLLMfp4
150 tok/s20,866$0.023vLLMfp4
200 tok/s15,470$0.031vLLMfp4

What serving actually costs

Converting the 50 tokens/s per user operating point to $ per million total tokens across rental pricing tiers from the SemiAnalysis AI Cloud TCO model.

Pricing tier$ / GPU / hr$ / 1M tokens
Hyperscaler$1.73$0.011
Neocloud$2.07$0.013
Retail$2.60$0.016

Frequently asked questions

How fast is gpt-oss-120b on B200?
At an interactivity target of 50 tokens/s per user on a single-turn chat workload (8k input / 1k output), B200 sustains 44,489 tokens/s per GPU serving gpt-oss-120b with vLLM in FP4. Peak measured throughput across all configs is 65,459 tokens/s per GPU.
How much does it cost to serve gpt-oss-120b on B200?
$0.011 per million total tokens at hyperscaler $/GPU/hr pricing, at 50 tokens/s per user. Neocloud and retail rental tiers are tabulated above; slower interactivity targets lower the cost further.
Which serving engines run gpt-oss-120b on B200?
The runs behind this page used TRTLLM, vLLM in FP4. Engines are rebuilt and re-benchmarked continuously, so the best config can change between visits.
How are these gpt-oss-120b numbers measured?
Every number is measured on real B200 hardware by the InferenceX fleet, sweeping concurrency on a single-turn chat workload (8k input / 1k output) to trace the throughput-versus-interactivity frontier; the newest run landed on 2026-05-31. The same derivation powers the InferenceX overview leaderboard.

Explore the data