With the recent rollout of Gemini 4 Neon replacing Argon in the Gemini app, I wanted to share some honest feedback—and see if others are feeling the same way.
While I understand the push for a cleaner model lineup, faster responses, and better compute efficiency, losing Argon from the Gemini app feels like a real step backward. In day-to-day use, Argon had a distinct edge: its answers were consistently deeper, more detailed, and far more meticulous than both Neon and Carbon.
Neon is quick, sure, but it tends to summarize too aggressively and skim over nuances. Even Carbon, despite being positioned for heavy-duty tasks, can sometimes gloss over the finer details that Argon used to catch naturally. For anyone doing thorough analysis or detail-sensitive work, that difference is immediately noticeable.
Google really needs to reconsider this trade-off. Optimizing for speed and efficiency shouldn’t mean sacrificing thoroughness—getting one thing at the expense of another doesn’t feel like an upgrade when the thing being lost is exactly what a large group of users rely on.
Hopefully, the team will consider either bringing Argon back as an option in the Gemini app, or tuning Neon and Carbon so they can actually match Argon’s level of depth and care.
Is anyone else missing Argon’s thoroughness since the switch?
I can totally agree, it would be great if Argon remained within the Gemini app, I love how it handles creative writing. Plus it’s great knowing I have that option
Please bring back Gemini 4 Argon to the Gemini app. I use Gemini for creative writing and literature, and the new Neon and Carbon models are completely unsuitable for this. They lack the literary depth, nuance, emotional resonance, and long-context consistency that Argon had. Replacing Argon has disrupted my ongoing writing projects in the app. Please restore access to Gemini 4 Argon in the model selector.
Thank you @KeinaLG, @Wanith_Gunawickrama, and @Prince_Uche for sharing your experiences. Seeing these responses confirms this isn’t an isolated impression—it reflects a real, practical regression that standard benchmarks don’t fully capture.
To give credit where it is genuinely due: since Gemini 4 Argon was introduced, the leap in reasoning, analytical thinking, and computational accuracy over earlier Pro and Flash generations has been truly remarkable. It was a praiseworthy breakthrough—so much so that it seemed to be the very reason Google replaced the dedicated Deep Think mode on Ultra with Argon.
That is precisely why its abrupt removal feels like such a setback. While I understand the goal of streamlining the model lineup and optimizing for speed, neither Neon nor Carbon serves as an adequate replacement for Gemini 4 Argon in my daily workflows, for three specific reasons:
Strict Prompt Adherence: Both Neon and Carbon frequently overlook or gloss over granular instructions in complex, multi-constraint prompts. Argon consistently follows detailed requirements faithfully without dropping edge cases or cutting corners.
Literary Depth & Natural Argumentation: Echoing @KeinaLG and @Prince_Uche, Argon strikes a rare balance of analytical depth, expressive flexibility, and natural, human-like prose. Neon optimizes for speed at the expense of accuracy and thoroughness (as @Wanith_Gunawickrama noted), while Carbon tends to feel noticeably more rigid on nuanced writing and argumentation.
Document Synthesis & Citations (Gemini Notebook): The difference is especially clear in document-grounded workflows like Gemini Notebook. Argon delivers far more detailed, precise, and deeply contextualized citations—and more importantly, it draws genuinely sharp, insightful arguments from source materials rather than settling for surface-level summaries.
For Gemini Ultra subscribers in particular, the core value of paying for the top tier lies in having consistent, reliable access to the deepest models for serious, ongoing projects. Removing a foundational model like Argon outright—without keeping an optional fallback—disrupts established workflows and noticeably weakens that value proposition.
Constructive request to the Gemini team: Please consider restoring Gemini 4 Argon to the Gemini app model selector, even as an optional or legacy choice for Pro and Ultra subscribers. Giving paying users the flexibility to prioritize instruction fidelity, citation accuracy, and analytical depth shouldn’t have to be sacrificed for faster default models.
I also wish to add my voice here. I loved Argon, I didn’t care it took 25% per question of the 5hr window, it was the most accurate LLM so far by a large mile, 0 hallucinations in my use and always thorough, left nothing to be desired.
Now its arGon, and this makes me feel shafted by Google. Why don’t they remove other .. like youtube premium light (useless anyway) but they remove Argon from this subscription is beyond my understanding.
I spent this morning wondering why they released Neon as it is obviously unusable, no issues at all with Argon, really good and sound. Regret to say will have to transfer to Grok as this morning has been painful with Neon, couldn’t even format a google docs file and made endless mistakes. took 10 attempts to rectify my document from all caps (that it introduced) and then nail in the coffin it right justified some paragraphs. I now have high blood pressure and less hair. Goodbye Gemini; helo my mate Grok
Honestly, you’re totally right. Neon doesn’t have the precise reasoning ability and natural flow that Argon possessed; I’ve noticed a drop in quality in my work when using Neon.
It’s annoying to see that this change was abrupt and left no way to go back to using Argon. In my particular case, I didn’t mind that Argon took longer because its responses were more thorough, rigorous, and explained everything in greater detail.
I sincerely hope that Argon becomes available as an intelligence option again.
i am following a law degree , when i used argon it was very help to summarized sections and articles very accurately in statues but now neon not do it properly always made mistakes, i am very disappointed about neon my request is to re introduce argon again
For statistical usage, damn, I would say, it’s incredibly detailed and thorough when I told it to calculate and estimate based on the data I gave them (although sure, it’s only for optimizing game grinds, but damn, it even gave me detailed and optimized next steps on what upgrade to take)
Sure, it may take a damn long time to process, but the quality is incredible.
Fully agree with these posts, Argon was phenomenal for deep research, particularly deep esoteric research which is what I used it for. The ultra low hallucination rate combined with the open-mindedness of the model was making for mindblowing discoveries that all fact-checked true. Neon can’t compare and is doing alot more default safety tuning behavior, dismissing alot of the subject matter I ask it about.
Please allow us to use Argon! I don’t mind the high usage either, was worth it.
Totally agree with OP. Replacing 3.1 Pro and 4 Argon with 4 Neon in the $20 AI Pro tier feels like a massive downgrade in real-world usage, regardless of what synthetic benchmarks say.
Neon feels overly distilled and lazy on complex UI generation and nuanced reasoning, even free 4 Hydrogen often produces more complete and thoughtful output than Neon. Removing actual depth and context handling while keeping opaque, unpredictable rate limits completely kills the value of the $20 subscription.
For $20/month, competitors give access to their actual flagship models without paywalling intelligence behind a $100 tier. Many of us are more than willing to pay full price directly for a pure AI subscription (even without the extra Google Drive storage bundle), if we actually get a smart, deep model like 3.1 Pro or Argon back. Right now, there is simply nothing worth paying for in AI Pro.
I didn’t want critize you for this decision, Google. But please, give us a bit long time for using 4 Argon and 3.1 Pro models again. Many peoples can’t afford upgraded to Ultra plan only to get 4 Carbon model for just maintained the coversation quality.
But seriously, i’ve been spending long time using Gemini app and subscribed Pro Plan for a year and never bending to another AI apps because you have great models to make an creative writings and story. I prioritize the depth of story than speed of story can be completed.
Think again about non-programmer or developer Gemini users, Google.
I hope you read this topics and know what we feel too, Google.
I was a heavy user of 4 Argon as well, though I feel another model is being overlooked here too: 3.1 Pro.
Personally, I had a great workflow combining both—I used 4 Argon to structure, organize, and wrap up my writing and stories, while 3.1 Pro was my go-to for deep discussions and subtle creative nuances. Honestly, 3.1 Pro was the best model I’ve ever used. It hit a sweet spot: it had solid accuracy, didn’t burn through context/tokens as heavily as Argon or Carbon did before the update, and most importantly, its prose felt genuinely creative and human.
These newer replacements feel like they are drifting back toward a more robotic, mechanical tone, losing the human warmth and emotional nuance I used to get from 4 Argon and 3.1 Pro.
Please consider bringing them back to the app, or at least giving creators an option to access them.
I am absolutely furious.
Just like that, Argon AND 3.1 Pro are gone - replaced by some watered down model made up out of thin air as it appears to me as regular user.
Argon was amazing and I want it back.
I give Google 30 days to bring Argon back or something with similar capabilities for in depth natural science subjects and latest developments in medical science.
Else I’ll cancel my Gooogle AI Pro subscription and have a look at the competition.
I was quite satisfied with 3.1 Pro, it was really good for the time but now it’s showing its age.
For me Argon is an excellent successor and it really was about time Google came up with a 3.1 Pro successor, it took them long enough.
Now suddenly, like literally within the time of a browser refresh, Argon is gone. Just gone. And 3.1 Pro on top. I am pissed.
So annoyed with it gone too! I’ve been into creative writing for quite a while now, and when Argon came out, I was so happy with the way it edited my work, and with the prompts it gave me. Now it’s already gone? It was the most in-depth one they have done. It feels like a huge downgrade with what they have left us. They are really terrible and completely already changed some of my current work, with how I wanted it. It forget things I’ve just gave it, and the responses are hugely lacking and has less depth. Very disappointed, with Argon and Pro gone. I’m literally wondering what I’m paying for when there is literally no Pro model for the £18.99 subscription? It genuinely feels like I’m using a free model now but still paying for Pro prices?
Yeah. Why remove both 3.1 Pro and Argon? By doing this you have indirectly said that Google subscriptions are just for coding tasks. Neon is works at all of the other jobs as argon. Infact it feels like it’s a distilled low parameters model made from carbon. And why don’t you also test your models on creative writing benchmarks?