Model Overview
GPT-4.1 is an API model released by OpenAI on April 14, 2025, focused on coding, instruction following, and long-context understanding. Its API model ID is gpt-4.1, and its corresponding release snapshot is gpt-4.1-2025-04-14. (openai.com)
Capabilities and Uses
- Supports text input and output and image input; supports function calling, structured outputs, and streaming. Suitable for applications including code generation and modification, long-document analysis, image understanding, and tool calling. (openai.com)
- The context window is 1,047,576 tokens, and the maximum output is 32,768 tokens. The knowledge cutoff date is June 1, 2024. (developers.openai.com)
- At release, OpenAI reported a score of 54.6% on the SWE-bench Verified coding benchmark; this figure is a specific evaluation result and does not guarantee performance on real-world tasks. (openai.com)
Limitations and Status
- Does not support audio or video input or output; images are supported as input only. Retrieval or reasoning errors may still occur in long-context tasks, and actual results depend on the task and input. (developers.openai.com)
- In OpenAI's initial testing, time to first token was approximately 15 seconds for a 128,000-token context and approximately 1 minute for a 1 million-token context; this is not a latency guarantee for all requests. (openai.com)
- As of November 18, 2026, OpenAI had marked the snapshot as deprecated, with shutdown scheduled for January 14, 2027; OpenAI's recommended replacement model is
gpt-6-luna. (developers.openai.com)
Pricing
As of November 18, 2026, the standard text token prices listed on OpenAI's official API model page are billed in US dollars (USD) per 1 million tokens: $2.00 for input, $0.50 for cached input, and $8.00 for output. The cached input price applies only to cache-eligible inputs; other platforms, fine-tuned models, and tool calls may be billed differently. (developers.openai.com)