Open Models Almanac
Methodsinference & servinginference quantization

specific method · filed under inference & serving

BF16 inference

Inference performed with weights and activations in BF16, including an implementation that upcasts weights before execution.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See inference quantization for the methods filed beside it.

Also called bfloat16 inference.

sources
2
models
2
labs adopt it
2
strongest
default

Documented in

OpenVLA project (Stanford et al.)

Evidence

2 quotes from 2 sources, one card per model, each opening with its strongest.

How sources treat it
used 1default 1

gpt-oss-120b

OpenAIadopted1 source

In this implementation, we upcast all weights to BF16 and run the model in BF16.

usedsoftware implementationin gpt-oss reference PyTorch implementationOpenAI
↑