MDL
gpt-oss-20b
openai
Verified against Artificial Analysis · Jul 13, 2026
Parameters
20B
Downloads
7.3M
on HF
Likes
4.8K
Avg score
52.1
10 benchmarks
AA Intelligence
14.9
index
Speed
273 tok/s
TTFT 0.4s
Released
Aug 4, 2025
About
gpt-oss-20b is a 20B-parameter open-weights model in the gpt family from openai. Benchmarked on 10 evals, averaging 52.1.
apache-2.0text-generation
Benchmark scores · 10
avg 52.1 — higher is betterThe research behind this model
all papers mentioning it →gpt-oss-120b & gpt-oss-20b Model CardBenchEvolver: Frontier Task Synthesis via Solution-Centric Evolution▲ 5 on HF · May 31, 2026The Chain Holds, the Answer Folds: Trace-Answer Dissociation in Reasoning Models Under Adversarial Pressure▲ 1 on HF · May 27, 2026Pruning and Distilling Mixture-of-Experts into Dense Language Models▲ 1 on HF · May 27, 2026Hurwitz Quaternion Multiplicative Quantization for KV Cache CompressionMay 26, 2026ReacTOD: Bounded Neuro-Symbolic Agentic NLU for Zero-Shot Dialogue State TrackingMay 18, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed273 tok/s
Latency (TTFT)0.4s
First answer7.7s
Blended price$0.088 / 1M
Coding index20.7
Math index89.3
Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$0.2/M