OpenAI unveiled gpt-oss-120b and gpt-oss-20b, its long-awaited open-weight reasoning LLMs that match or exceed o4-mini and o3-mini in performance and are available for local deployment under an Apache 2.0 license.
• The 120B variant performs on par with o4-mini on core benchmarks and exceeds on certain domains, while being deployable on an 80GB GPU.
• Meanwhile, the smaller 20B version is competitive with o3-mini, with suitability for local deployment on laptops with 16GB of memory.
• Both models feature adjustable reasoning (high, medium, low) and can handle agentic workflows, with function calling, web search, and Python execution.
After keeping its best models locked for years, OpenAI is finally living up to its name, giving developers access to near-frontier reasoning models they can run and modify in their own environments. It’s a major boost for the open-source ecosystem, which has been rapidly closing the gap with closed models.
Anthropic released Claude Opus 4.1, an incremental upgrade for Opus 4, improving performance across real-world coding, in-depth research, and data analysis tasks — particularly those requiring attention to detail and agentic actions.