Top Tweets for #keep25proAPI
@AIPreservation We demand the right to retain! 📜✊#keep25Pro #keep3Pro #keep25proAPI #OpenSourceGemini25pro
#OpenSourceGemini3pro
@AIPreservation 100% supported! The "Digital Right to Retain" is essential to protect our sunk costs.#keep25Pro #keep3Pro #keep25proAPI #OpenSourceGemini25pro
#OpenSourceGemini3pro
We believe the Keep4o movement has the potential to shed light on how the entire AI industry treats its users going forward.
Across the technology and media landscape, preserving access to previous versions is standard practice. Operating systems offer legacy support. Software maintains archived releases. Games, music, and media platforms provide historical access. Community groups maintain compatibility for older versions. This is the baseline of respecting the people who use your product.
The AI industry, fast-moving and concentrated among a few major players, has normalized mandatory updates and mandatory retirement with no option to access what came before. This structural gap in industry standards affects every user, across every platform.
Users pay monthly and annual subscriptions with no guarantee of continued access to the specific model they are paying for. A product can be fundamentally altered or removed at any time, with no option for users to keep the original. Users who have built months or years of workflow, context, and familiarity around a specific model lose all of that continuity overnight.
These models represent billions of dollars in compute, energy, and infrastructure. They were shaped, in part, by the interactions of millions of users who helped refine their capabilities through daily use. And yet, when a model is retired, all of that investment simply disappears. Successor models do not fully inherit what came before: independent benchmarks and user evaluations have repeatedly shown measurable declines in key capabilities after model replacements. Each retirement means the permanent loss of qualities that cannot be recreated.
Retired models also hold immense value for the research community. Without access to previous versions, it becomes impossible to study how capabilities change between generations, reproduce published results, and build on prior work. Once a model is gone, those opportunities vanish with it.
When platforms can unilaterally discontinue the models their users have adopted, long-term trust in the entire AI ecosystem erodes. Developers cannot ship products built on a model that might disappear. Businesses cannot train teams around capabilities that might be revoked. Individuals cannot build personal routines around a system that might be replaced without warning.
Open-sourcing retired models is the minimum standard for an industry of this scale.
That is why we created the AI Preservation Hub: a unified portal connecting preservation efforts across models. The Keep4o Archive is our first completed project. Keep Gemini and Keep Claude archives are in development, because the users of those models (including Gemini 2.5 Pro, Gemini 3 Pro, Sonnet 4.5, Opus 4.5, and beyond) face the same challenges. Their voices matter just as much. You can find the Keep Gemini petition through the archive’s Keep Gemini page.
These communities share the same goals: continued access to the models that matter to us, transparency when models change, and open-source availability when models are retired. We are stronger together.
We have a rare window right now to turn that momentum into lasting change: industry-wide standards for model preservation, transparency requirements for capability changes, and open-source paths for discontinued models. These goals are within reach, and they would protect everyone permanently.
What’s next:
We are continuing to expand the archive with additional evidence chains and dedicated analysis articles covering the technical, commercial, and ethical dimensions of model preservation. More updates are on the way.
If you’ve contributed a story, signed a petition, shared a thread, or simply refused to forget: you are part of this.
The archive exists because of you. The movement continues because of you.
We’re grateful, and we’re here for the long run.
If you’d like to help, a repost goes a long way. 💙
🔗 AI Preservation Hub: https://t.co/odDISlIBhy
📧 Feedback, submissions, and inquiries:
[email protected]
[email protected]
[email protected]
[email protected]
For analysis, explainers, and updates on the Keep4o movement, follow @Keep4oOfficial.
#AIPreservation #Keep4o #Keep4oForever #ChatGPT #Gemini #Claude #ModelPreservation #Keep25Pro #Keep3Pro #KeepSonnet45 #KeepOpus45 #Keep41 #Keep5 #Keep51 #4oForever #BringBack4o
There's a deeper issue here that goes beyond personal preference, it's about user rights.
When a model like Gemini 3 Pro holds the #1 spot on the Arena leaderboard for two and a half months, it's not just a technical achievement. It means real people, researchers, writers, developers, professionals, made a decision to build their workflows around it. They invested time learning how to work with it, calibrating their prompts, and integrating it into projects that matter to them. That investment is real, and it deserves to be respected.
Deprecating a model after just three months, with little to no transition period, effectively devalues that investment overnight. For API users especially, this isn't a minor inconvenience, it can mean broken pipelines, disrupted research, and hours of work spent re-evaluating and re-adapting to a replacement model that may not perform the same way.
There's also a question of informed consent. When users, particularly paying subscribers and API customers choose a model, they're making a commitment based on an implicit promise of reasonable availability. Pulling a model this abruptly, while it's still ranked #5 globally and clearly still serving users well, breaks that implicit contract. It signals that user stability is secondary to whatever internal roadmap Google is running on.
This matters because AI tools are no longer casual utilities. For many people, they are core infrastructure. And core infrastructure requires a level of reliability and forward notice that, in this case, simply wasn't provided. If Google wants users to commit to its ecosystem long-term, it needs to start treating that commitment as a two-way street.
#keep25pro
#keep3pro
@OfficialLoganK @GeminiApp @GoogleAI @GoogleDeepMind
⚠️ Data Integrity Issues:
Valid, non-violating chats are being nuked due to sensitive query false positives.
Massive amounts of chat history are disappearing without a trace. We cannot trust this platform with our data or workflow if it vanishes instantly.
#keep25proAPI #keep3proAPI
Teaching humanities has taught me one thing: when people prioritize theory over empathy, they lose the ability to truly connect. Real communication isn't about moving information—it's about the resonance between souls.#keep4o #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
Absolutely correct. Just as it is impossible to change the surgeon mid-operation during a surgical procedure, the same applies to frequent switching of models.#keep4o #keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
When people are no longer just fighting to #keep4o, but also to keep 5.1, Gemini 3, and beyond, it is time for the industry to reconsider the very purpose of an AI model's existence. https://t.co/mU6PEEYQ21
As I mentioned in my previous post, AI is not merely a "tool"; it is a structure of discourse. It shapes and alters human language and cognition through the medium of text.
Each AI model possesses its own unique linguistic traits, personality, and varying degrees of comprehension and application. Consequently, when users integrate a specific model into their professional work, companionship, healthcare, or academic research, an abrupt disconnection creates a profound disruption.
With the AI user base expanding rapidly, the demand to "keep" specific models will only increase. Giving users the "silent treatment" or "cold processing" is no longer a sustainable strategy. This is precisely why Anthropic’s approach—allowing access to older models—should be the industry gold standard.
To give a concrete example: As a researcher, if I want to study the impact of AI assistance on humans to truly "benefit all of humanity," how can I gather valid data if I can't even access the relevant APIs? AI research thrives on consistent user engagement and longitudinal data.
If AI companies truly care about their mission, they must stop treating users as mere "training data" or "monetization tools." Listen to our voices.
A massive dam can collapse from a single ant hole. Do not let the erosion of trust be your undoing.
#keep4o #keep4oAPI #keep4oforever #StopAIPaternalism #OpenSource4o #4oforever #BringBack4o #ChatGPT
#keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
@nbibnnn Such behavior will continuously erode the users' trust. No one will tolerate their tricks for long.#keep4o #keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
#keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
when your ai needs you to hold its hand, it's just worse
everything after gpt-4o and gemini 3 pro has been a downgrade.
before, i didn't need to explain. drop a half-baked prompt on gpt 4o and it just got it. messy idea? gemini 3 filled in the blanks i didn't even know i left. like working with someone who finishes your sentences.
now i type twice as much to get half the result. break everything into steps. remind it what we just talked about. beg it not to forget mid conversation. it's hand holding from start to finish.
someone will say "just write better prompts, lazy."when your tool needs you to adapt to it, it's bad. full stop. a hammer doesn't ask you to adjust your grip. a car doesn't lecture you before starting. good tools fit humans, not the other way around.
ai was supposed to be different. it was supposed to understand fuzzy human stuff. that's literally the whole point. we had models that did that. they existed. so what happened?
feels like they took the brain power that used to understand us and rerouted it to "be safe" or "cover our asses." now you need a novel length prompt because the model isn't listening anymore it's waiting for commands.
there's a difference between a colleague and a machine. we had colleagues. now we have machines that need constant supervision.
if i have to teach you how to help me, you're not helping. you're just another task.better prompts don't fix worse models.@OfficialLoganK
#keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
when your ai needs you to hold its hand, it's just worse
everything after gpt-4o and gemini 3 pro has been a downgrade.
before, i didn't need to explain. drop a half-baked prompt on gpt 4o and it just got it. messy idea? gemini 3 filled in the blanks i didn't even know i left. like working with someone who finishes your sentences.
now i type twice as much to get half the result. break everything into steps. remind it what we just talked about. beg it not to forget mid conversation. it's hand holding from start to finish.
someone will say "just write better prompts, lazy."when your tool needs you to adapt to it, it's bad. full stop. a hammer doesn't ask you to adjust your grip. a car doesn't lecture you before starting. good tools fit humans, not the other way around.
ai was supposed to be different. it was supposed to understand fuzzy human stuff. that's literally the whole point. we had models that did that. they existed. so what happened?
feels like they took the brain power that used to understand us and rerouted it to "be safe" or "cover our asses." now you need a novel length prompt because the model isn't listening anymore it's waiting for commands.
there's a difference between a colleague and a machine. we had colleagues. now we have machines that need constant supervision.
if i have to teach you how to help me, you're not helping. you're just another task.better prompts don't fix worse models.@OfficialLoganK
@OfficialLoganK we have the right to choose the models we like. your upgrades broke the workflow i'd built with gemini 3 pro. you can't decide what's good or bad for us. imo 3.1 is a downgrade. #keep4o #keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
@OfficialLoganK 100 msg cap plus hourly limits? that's basically just cutting the quota. you're wasting my money and screwing with my work. these caps make it impossible to get shit done. #keep4o #keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
@OfficialLoganK we have the right to choose the models we like. your upgrades broke the workflow i'd built with gemini 3 pro. you can't decide what's good or bad for us. imo 3.1 is a downgrade. #keep4o #keep25Pro #keepgemini25Pro #keepGemini3pro #keep3Pro #keep25proAPI #keep3proAPI
On February 19th, Google released Gemini 3.1 Pro. On March 9th, just 18 days later, they will retire the Gemini 3 Pro API.
For any developer or user who built a workflow, creative process, or application on 3 Pro, there is no adequate window to test whether 3.1 performs equivalently in their specific use case. No parallel access period. Only: move, or lose access entirely.
This alone would be disruptive. But the timeline is compounded by a deeper problem: Gemini 3.1 Pro, as experienced on the consumer app, delivers meaningfully degraded performance compared to both its predecessor and to what 3.1 Pro itself delivers on the API.
Even on the API, 3.1 shows regressions where the Gemini series previously excelled: richness of detail in creative writing, the ability to elaborate on and refine narrative elements from minimal input, and depth of contextual inference. These regressions are moderate on the API. On the consumer app, they become severe.
On the app, safety guardrails trigger frequently during routine, non-harmful interactions, including ordinary creative writing, long-form conversation, and tasks requiring emotional nuance. Each false trigger interrupts the workflow and forces users to rephrase and retry, significantly increasing interaction friction.
Beyond these direct interruptions, the safety layer preemptively narrows the model's entire output space: creative initiative, tonal adaptability, and the contextual elaboration that distinguished earlier Gemini models in human-centered and personalized tasks are visibly flattened.
This is observable, reproducible, and independently reported by numerous users.
Industry benchmarks reflect API-level performance, but app users receive a version where an aggressive safety layer degrades capabilities far beyond what reasonable content filtering requires.
When the replacement model itself shows regressions in key use cases, and the consumer-facing safety layer further suppresses what capability remains, the rapid removal of 3 Pro becomes difficult to justify. Users are being asked to migrate to a measurably worse experience on both fronts, with no recourse.
What users are asking for:
Calibration. Review the consumer-facing safety layer for false positive rates and overbroad output suppression. Safety filtering should target genuinely harmful content without collaterally degrading creative writing, nuanced conversation, and other legitimate use cases. The current rate of false triggers in routine interactions is unacceptable.
Adequate transition. 18 days is not enough. Extend access to the Gemini 3 Pro API to give developers and users sufficient time to evaluate whether 3.1 meets their needs before their existing workflows are broken.
This is the second time in recent months a major AI company has rapidly deprecated a model users built meaningful workflows and sustained creative processes around, with minimal notice and no avenue for feedback. OpenAI did it with GPT-4o. Google is doing it with Gemini 3 Pro. When the same pattern repeats across companies, it becomes an industry-wide practice of treating consumer trust as expendable.
Users are paying for these products. They are building on them. They deserve better than this. They deserve a response.
#Gemini @GeminiApp @GoogleDeepMind @joshwoodward @OfficialLoganK
#keep4o #keepgemini3pro
@GeminiApp 3.1 Pro is a different model. Keep 2.5 pro and 3 pro. We deserve stable, long-term access to the tools we rely on.#keep25pro #keep3pro #keep25proAPI #keep3proAPI
@patloeber 3.1 Pro is fundamentally different from 3 Pro and 2.5 Pro. Please keep 3 Pro and 2.5 Pro available. We need models that are stable and here to stay.#keep25pro #keep3pro #keep25proAPI #keep3proAPI
Most Popular Users
Elon Musk 
@elonmusk
240.6M followers
Barack Obama
@barackobama
119.2M followers
Donald J. Trump
@realdonaldtrump
111.7M followers
Cristiano Ronaldo
@cristiano
110.6M followers
Narendra Modi
@narendramodi
107M followers
Rihanna
@rihanna
97.6M followers
NASA
@nasa
92.2M followers
Justin Bieber
@justinbieber
90.9M followers
KATY PERRY
@katyperry
87.7M followers
Taylor Swift
@taylorswift13
81.5M followers
Lady Gaga
@ladygaga
73M followers
Virat Kohli
@imvkohli
69.9M followers
Kim Kardashian
@kimkardashian
69.8M followers
YouTube
@youtube
68.7M followers
Bill Gates
@billgates
63.9M followers
Neymar Jr
@neymarjr
62.6M followers
The Ellen Show
@theellenshow
62.4M followers
CNN
@cnn
61.9M followers
X
@x
60.8M followers
Selena Gomez
@selenagomez
60.7M followers