OpenAI has released GPT-4.1 and GPT-4.1 Mini in ChatGPT, enhancing performance in coding, instruction-following, and long-context tasks. First launched via API in April, these models are now integrated into ChatGPT (as of May 14, 2025). Both come with a June 2024 knowledge cutoff and support up to 1 million tokens of context.

Highlights:
• Coding: GPT-4.1 scored 54.6% on SWE-bench Verified—up 21.4 points from GPT-4o.
• Instructions: Achieved 38.3% on Scale’s MultiChallenge—10.5 points higher than GPT-4o.
• Context: Leads on Video-MME with 72.0% on long-context tests.
• Efficiency: GPT-4.1 Mini reduces latency by half and costs 83% less than GPT-4o. GPT-4.1 Nano (API-only) is the fastest and cheapest model.

Models and Comparisons

GPT-4.1 improves upon GPT-4o, keeping its multimodal abilities while boosting accuracy in coding and instructions. GPT-4.1 Mini and Nano succeed GPT-4o Mini with lower costs and faster response times. Compared to the original GPT-4 (Sep 2021), GPT-4.1 adds better instruction handling, faster inference, and longer context limits.

While direct comparisons to rivals like Gemini aren't provided by OpenAI, analysts note GPT-4.1 is clearly designed to compete with leading models from Google and Anthropic.

"GPT-4.1 is now available in ChatGPT for Plus, Pro, and Team users... it excels at coding with stronger instruction-following than GPT-4o," noted AI engineer Tibor Blaho. He emphasized that GPT-4.1 Mini delivers “significant improvements” in intelligence and usability.

Availability in ChatGPT

As of May 14, GPT-4.1 is available to Plus, Pro, and Team users. Free-tier users will see GPT-4.1 Mini as their fallback model once GPT-4o usage is exhausted. Enterprise and educational access is expected soon.

Pricing and limits remain unchanged. GPT-4.1 and Mini follow the same quota and billing as GPT-4o models. No new charges apply across ChatGPT plans. The models are also rolling out in Azure OpenAI and GitHub Copilot.

Expert Commentary

OpenAI says the models were optimized for real-world use, especially in software development and automation. Sam Altman described them as offering "better models at faster speeds" with greater cost efficiency. AI analyst Nathan Lambert called GPT-4.1 a leap in “intelligence per dollar.”

Early reports and benchmarks support these claims, especially for tasks involving code, reasoning, and long content. Community feedback and testing continue, but GPT-4.1 appears to restore OpenAI’s edge in multiple use cases.

Overall, GPT-4.1 sets a new standard for ChatGPT, refining multimodal AI with sharper coding skills, better instructions, and faster, cheaper access. Its broad rollout suggests OpenAI is confident in its ability to compete at the highest level.