OpenAI has unveiled the GPT-4.1 model, an LLM model with three sizes that enhance the capabilities of the GPT-4o, allowing it to support inputs of up to 1 million tokens and improving programming writing abilities surpassing the OpenAI o3-mini model, albeit at a significantly higher cost.
During testing using SWE Bench Verified, OpenAI showcased the GPT-4.1 with an impressive score of 55% (52% excluding problematic test cases), demonstrating accurate responses for diff for file corrections. The key enhancement lies in expanding the input capacity to 1 million tokens, enabling the model to effectively locate information in the needle in haystack test.
The model is available in three versions: gpt-4.1 priced at $2 per million tokens for input and $8 per million tokens for output, gpt-4.1-mini priced at $0.4 per million tokens for input and $1.6 per million tokens for output, and gpt-4.1-nano, a smaller, faster model priced at $0.1 per million tokens for input and $0.4 per million tokens for output, offering a 20% cost savings compared to GPT-4o.
OpenAI is now offering the GPT-4.1 exclusively through API, with sales directly from OpenAI and through Azure AI services.
TLDR: OpenAI introduces the GPT-4.1 model with enhanced features, improved testing scores, and varying pricing options for different versions, available exclusively through API services provided by OpenAI and Azure AI.
Leave a Comment