OpenAI unveiled its latest GPT-4.1 series: GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano yesterday, and we have already integrated all those models seamlessly into our platform's OpenAI modifiers so that you can take advantage of them straight away.
This release delivers substantial improvements in coding, instruction following, and long-context processing, offering unparalleled cost efficiency and speed, establishing it as the most cost-effective and high-performing model available exclusively via the OpenAI API.
Summary of the enhancements:
- Superior Coding Capabilities
GPT-4.1 excels in real-world coding tasks, scoring 54.6% on SWE-bench Verified, a 21.4% improvement over GPT-4o. It handles complex software engineering challenges, produces reliable code diffs, and minimizes extraneous edits (down to 2% from 9%). For front-end development, it creates more functional and visually appealing web apps, preferred 80% of the time over GPT-4o. - Enhanced Instruction Following
The series shines in following complex instructions, achieving 38.3% on Scale’s MultiChallenge benchmark (10.5% better than GPT-4o) and 87.4% on IFEval. This reliability powers robust agentic systems, enabling tasks like automated customer support or data extraction with minimal errors. Our modifiers make it easy to customise these capabilities for specific use cases, ensuring consistent outputs. - Massive Context Window
All GPT-4.1 models support a 1-million-token context window - equivalent to eight React codebases - ideal for processing large documents or code repositories. They excel at retrieving and reasoning across long contexts, scoring 72.0% on Video-MME for long video understanding and maintaining accuracy in multi-hop tasks like Graphwalks (61.7%). Our platform’s modifiers allow users to tap into this for legal analysis or codebase navigation applications. - Cost efficiency and speed
- GPT-4.1: Priced at $2.00 per 1M input tokens and $8.00 per 1M output tokens, it’s 26% cheaper than GPT-4o for median queries. With 75% prompt caching discounts, costs drop further, making it economical for high-volume tasks.
- GPT-4.1 mini: At $0.40 per 1M input tokens, it cuts costs by 83% compared to GPT-4o while halving latency, matching or surpassing GPT-4o in intelligence evals. Perfect for low-latency needs like real-time analytics.
- GPT-4.1 nano: The fastest and cheapest at $0.10 per 1M input tokens, it delivers exceptional performance for lightweight tasks like classification, with first-token latency under five seconds for 128,000 tokens.
By supporting GPT-4.1 models in our modifiers, we empower users to build sophisticated, cost-effective AI solutions tailored to their needs. Whether automating code reviews, processing vast datasets, or creating responsive agents, these models deliver top-tier performance with lower costs and faster responses. The integration ensures flexibility, letting users fine-tune outputs as easily as tweaking a formula, unlocking new possibilities for innovation.
OpenAI’s GPT-4.1 series, now supercharged by our platform, sets a new standard for AI-driven productivity. Explore these models today and transform your workflows with precision and efficiency.