gpt-oss-20b

Gpt family · 21.0B params

Evidence: 2 sourced runs · 1 awaiting review · 1 independent source · first recorded Aug 17, 2026 · last recorded Aug 24, 2026. What these labels mean.

What the gpt-oss-20b numbers mean

The fastest verified result on our board is 93.4 output tok/s on an RTX 3060 x2 at Q4_K_M with a 65536-token context. That is about 10.7 ms per generated token — 500 tokens takes roughly 5.4 seconds. Prompt processing on the same record runs at 895.4 tok/s, so a 2,000-token prompt is read in about 2.2 seconds. Peak VRAM was measured at 12.2 GB. Across the 3 records we hold for this model the range is 57.1–307.2 tok/s on 3 GPUs, because quantisation, context length and the engine build all move the number.

How much VRAM does gpt-oss-20b need?

gpt-oss-20b has 21.0B parameters. These estimates use the same formula as the VRAM calculator: weights, plus about 15% for a moderate-context KV cache and runtime overhead, plus a small engine floor.

Long contexts and large batches raise the real number — read the smallest card as "possible with a reduced context". Measured throughput is on the benchmark board.

Models with a similar footprint

Ranked by distance from gpt-oss-20b's parameter count. Full benchmark board.

All benchmarks Methodology