OpenAI drops first open-weight models since GPT-2

OpenAI released gpt-oss-120b and gpt-oss-20b under Apache 2.0 license. First open language models from OpenAI in 5+ years.

• gpt-oss-120b: 117B total params, 5.1B active per token, runs on single 80GB GPU
• gpt-oss-20b: 21B total params, 3.6B active, fits in 16GB memory
• Matches o4-mini on reasoning benchmarks
• Full chain-of-thought reasoning, tool use, 128k context
• Available on Hugging Face with quantized weights

https://openai.com/index/introducing-gpt-oss/

Why it matters: First-time founders can now run OpenAI-grade reasoning models locally without API costs.

Source: OpenAI official blog.
Brew your ideas with Startup Beaker.

📱 🔗Startup Beaker