OpenAI releases GPT-4.1 in ChatGPT for Plus, Pro, and Team subscribers “by popular request”, and replaces GPT-4o mini with GPT-4.1 mini for all ChatGPT users

OpenAI is releasing its GPT-4.1 and GPT-4.1 mini AI models in ChatGPT, the company announced in a post on X Wednesday.

TechCrunch 2025-05-15 Maxwell Zeff

Context & Ripple Effects

OpenAI had introduced the GPT-4.1 family through its API with emphasis on coding, instruction following, and long-context work; this moves that new model family from developer access into ChatGPT. It also supersedes the smaller GPT-4o variant that had itself replaced GPT-3.5 Turbo in the company’s lower-cost lineup.

The release continues a recent pattern of tiered ChatGPT distribution: paid users had received o3 and o4-mini, while a lightweight deep-research option was extended to free users. Model choice is becoming a recurring product-control mechanism rather than a one-time launch decision.

First-order effects

Plus, Pro, and Team subscribers gain GPT-4.1 as a ChatGPT option, while all users are shifted from GPT-4o mini to GPT-4.1 mini.
The change makes the GPT-4o mini replacement introduced in 2024 an explicit handoff to a newer small-model default, affecting the baseline experience for free users.

Second-order effects

Paid-plan differentiation now depends not only on feature access but also on which model users can select, reinforcing OpenAI’s segmentation across free and subscription tiers.
Developers and teams using both the API and ChatGPT can align interactive work with the GPT-4.1 models already available via the API, reducing the gap between OpenAI’s developer and consumer product surfaces.

Third-order effects

If these replacements continue, ChatGPT will increasingly operate as a managed model portfolio: users receive changing defaults while higher tiers retain more explicit model access.
That makes frontier-model access governance a core competitive lever, with providers balancing broad access to capable smaller models against premium access to newer or more specialized systems.

The trend: AI assistants are shifting from static flagship releases toward continuously managed, tiered model portfolios optimized for different user segments and workloads.

Discussion

@openai @openai on x
By popular request, GPT-4.1 will be available directly in ChatGPT starting today. GPT-4.1 is a specialized model that excels at coding tasks & instruction following. Because it's faster, it's a great alternative to OpenAI o3 & o4-mini for everyday coding needs.
@joheidecke Johannes Heidecke on x
3/ While this is a notable improvement, GPT-4.1 doesn't introduce new modalities or ways of interacting with the model, and doesn't surpass o3 in intelligence. This means that the safety considerations here, while substantial, are different from frontier models.
@michpokrass Michelle Pokrass on x
gpt-4.1 landing in chatgpt today!! we were initially planning on keeping this model api only but you all wanted it in chatgpt :) happy coding!
@kevinweil Kevin Weil on x
By popular request: ✨GPT 4.1✨ is now available in ChatGPT for Plus/Pro/Teams subscribers (and soon to Enterprise/Edu). We built it for developers, so it's very good at coding and instruction following—give it a try! Also, ✨GPT 4.1 mini✨ is replacing GPT 4o mini everywhere
@joheidecke Johannes Heidecke on x
2/ Before launching GPT-4.1 in the API, we ran evaluations to test the model's capabilities and safety. It excels at coding and instruction following - things that are extremely helpful for developers.
@gdb Greg Brockman on x
GPT-4.1 has been a surprise hit — now available in ChatGPT:
@kalomaze @kalomaze on x
why didn't they just do this from the beginning. did their “high taste testers” not test this one
@koltregaskes @koltregaskes on x
Massively frustrating, OpenAI released GPT-4.1 without the API variant's of 1M context window. @OpenAI, come on! The large context window is the biggest use case for the model and the reason I've been begging for it in ChatGPT. 😢 Hopefully we get a big context window (both in [im…
@joheidecke Johannes Heidecke on x
4/ GPT-4.1 builds on the safety work and mitigations developed for GPT-4o. Across our standard safety evaluations, GPT-4.1 performs at parity with GPT-4o, showing that improvements can be delivered without introducing new safety risks.
@davidondrej1 David Ondrej on x
right now, GPT 4.1 is my favorite model for building AI Agents. it just does what you want - consistently & predictably.
@slow_developer Haider on x
damn. this is huge.. gpt-4.1 is now available directly in chatgpt o3 for planning and gpt-4.1 for execution has been my favorite workflow so far gpt-4.1 is best at following instructions, making clean changes, and avoiding unnecessary code or comments good bye gpt-4o
@openaidevs @openaidevs on x
Coded up by GPT-4.1, rolling out today in ChatGPT. [video]
@kimmonismus @kimmonismus on x
Ok, somehow OpenAI made the model selection even more confusing. Not even I know when to use GPT-4o and when GPT-4.1. We have: - GPT-4o - GPT-04 mini - GPT-o4 mini high - GPT-4.1 - GPT-4.1 mini - GPT-4.5 - o3 Come on..
@btibor91 Tibor Blaho on x
Max tokens per user message in ChatGPT (Pro) for GPT-4.1 mini is 129,024 (though I only got it working with ~117k), and for GPT-4.1 it's 34,815 - max context length is apparently just ~128k tokens though? [image]
@elder_plinius @elder_plinius on x
lol gotta love when a new model is jailbroken out of the box by my memories 🤗 GPT-4.1 is a fun one [image]
@openai @openai on x
GPT-4.1 and GPT-4.1 mini underwent our standard safety evaluations. Detailed results are available in the newly launched Safety Evaluations Hub. https://openai.com/...
@diegocabezas01 Diego on x
I like GPT 4.1 [image]
@emollick Ethan Mollick on x
Not bad from GPT-4.1: “create something I can paste into p5js that will startle me with its cleverness in creating something that invokes the control panel of a starship in the distant future” First go, no errors. [video]
@joheidecke Johannes Heidecke on x
1/ Safety is core to every model we build at OpenAI. As we deploy GPT-4.1 into ChatGPT, we want to share some insights from our safety work. 🧵
@sal19 Sal Rodriguez on bluesky
Big one we've been working on for a while: AI research takes a backseat to profits as Silicon Valley prioritizes products over safety, experts say www.cnbc.com/2025/05/14/m... by @haydenfield.bsky.social, @jonathanvanian.bsky.social & @jennelias.bsky.social
@haydenfield Hayden Field on x
Five hours after our AI research story published — we reached out to all the AI labs 1-2 weeks beforehand — OpenAI pledged it would release safety test results more often & launched a webpage showing how its models score on tests for harmful content. https://www.cnbc.com/...
@openai @openai on x
Introducing the Safety Evaluations Hub—a resource to explore safety results for our models. While system cards share safety metrics at launch, the Hub will be updated periodically as part of our efforts to communicate proactively about safety. https://openai.com/...

Chronicles