Model Releases

OpenAI open-source reasoning model performance evaluation released

Original title:[Lmgame Bench] 🔥 OpenAI has just released two open‑weight reasoning models: gpt‑oss‑120B (~117 B) a...

AI summary

[Lmgame Bench] 🔥 OpenAI has just released two open-weight reasoning models: gpt-oss-120B (about 117 billion parameters) and gpt-oss-20B (about 21 billion parameters), the first open-weight OpenAI models since GPT-2. We tested both on Lmgame Bench across four interactive games: 🧱 Sokoban | 🟦 Tetris | 🔢 2048 | 🍬 Candy Crush. Here are their rankings (out of 25 points): → gpt-oss-120b → 12th place → gpt-oss-20b → 13th place

Key Takeaways

  • [Lmgame Bench] 🔥 OpenAI has just released two open-weight reasoning models: gpt-oss-120B (about…
  • We tested both on Lmgame Bench across four interactive games: 🧱 Sokoban | 🟦 Tetris | 🔢 2048 | 🍬…
  • Here are their rankings (out of 25 points): → gpt-oss-120b → 12th place → gpt-oss-20b → 13th place

Quick Facts

Published
Updated
Source
X:UCSD Hao AI Lab (@haoailab)
Category
Model Releases
AI score
81
Read the original Read the full article at the original source
Why it matters

OpenAI 终于开源了,这是 GPT-2 之后第一次放开权重,120B 和 20B 两个尺寸直接对标 Llama 和 Qwen 的开源生态。虽然游戏 benchmark 排名不算惊艳,但信号本身比分数重要得多,所有基于开源模型做产品的团队都得重新评估选型。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: OpenAI open-source reasoning model performance evaluation released. August 7, 2025. https://new.nat6.net/en/news/cmnxjn85o00h0sl9oz3hg57qx/

View all →