MDL
gpt-oss-120b
openai
Verified against Artificial Analysis · Jul 13, 2026
Parameters
120B
Downloads
4.4M
on HF
Likes
5K
Avg score
60.7
10 benchmarks
AA Intelligence
23.8
index
Speed
322 tok/s
TTFT 0.5s
Released
Aug 4, 2025
About
gpt-oss-120b is a 120B-parameter open-weights model in the gpt family from openai. Benchmarked on 10 evals, averaging 60.7.
apache-2.0text-generation
Benchmark scores · 10
avg 60.7 — higher is betterThe research behind this model
all papers mentioning it →gpt-oss-120b & gpt-oss-20b Model CardCompress-Distill: Reasoning Trace Compression for Efficient Knowledge Distillation▲ 2 on HF · Jun 4, 2026GRASP: Gated Regression-Aware Skill Proposer for Self-Improving LLM AgentsMay 28, 2026Metacognition as Reward: Reinforcing LLM Reasoning via Knowledge and Regulation SignalsMay 22, 2026Check Your LLM's Secret Dictionary! Five Lines of Code Reveal What Your LLM Learned (Including What It Shouldn't Have)May 21, 2026Effort as Ceiling, Not Dial: Reasoning Budget Does Not Modulate Cognitive Cost Alignment Between Humans and Large Reasoning ModelsMay 16, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed322 tok/s
Latency (TTFT)0.5s
First answer6.7s
Blended price$0.262 / 1M
Coding index30.4
Math index93.4
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$0.6/M