Will It Fit?
Estimate text inference VRAM across Transformers and vLLM with a compact, explainable breakdown.
Model pages and release notes are included here too.
GPT-OSS 120B · vLLM · Mixed MXFP4 + BF16 checkpoint
Required GPU VRAM (0.9 budget)
80.0 GB
Core estimate: 72.0 GB. Against RTX 4090 24GB, this leaves 54.2 GB of deficit.
At 4,096 tokens, the estimated max concurrency is 0 concurrent requests.
Quick read
Selected GPU
Runtime
Load dtype
GPU utilization
GPU count
KV cache dtype
Context length
Current concurrency
Max concurrency @ context
Class
Bandwidth
Nominal VRAM
Core estimate
Required GPU VRAM (0.9 budget)
Deficit
GPU compare
What fits this setup
Breakdown
Where the memory goes
Weights
117B resident parameters at 0.56 bytes each. Calibrated from the official checkpoint profile.
65.3 GB
KV cache
Concurrency 1, context 4,096, 18 dense layers @ 4,096 tokens + 18 sliding-window layers @ 128 tokens, 8 KV heads, BF16 cache storage.
0.2 GB
Runtime / safety overhead
Conservative buffer for allocator fragmentation, kernels, and runtime scratch space.
6.5 GB
Weights = parameter count × bytes per parameter.
KV cache grows with context length, KV-bearing layers, concurrent requests, and the selected KV cache dtype.
Show the substituted formulas
Weights
parameter count × bytes per weight
117B × 0.56 = 65.3 GB
OpenAI's GPT-OSS model card lists a 60.8 GiB checkpoint for gpt-oss-120b. The estimator uses that published mixed MXFP4 + BF16 resident checkpoint size directly.
KV cache
batch × effective KV tokens across attention layers × 2 × KV heads × head dim × bytes per KV element
1 × 76,032 × 2 × 8 × 64 × 2 = 0.2 GB
Dense attention layers keep the full 4,096-token cache, while sliding-window layers only keep 128 tokens. BF16 controls the bytes per stored KV element.
Overhead
max(1.5 GB, 10% of weights + KV cache + linear state)
max(1.5 GB, 10% of 65.4 GB) = 6.5 GB
This leaves room for runtime buffers instead of claiming an unrealistically exact fit.
Model
Selected model
GPT-OSS 120B
117B total • 5.1B active • 128,000 context • 8 KV heads
Total params
Active params
Layers
Hidden size
Attention heads
KV heads
KV-bearing layers