Today I asked Gemini to extract a table from an PDF appendix and create C++ data table with its contents. After 15 or so iterations with corrections and new mistakes, it eventually gave up. I was floored when it said “I’m sorry, I cannot do this simple task, I’ve exceeded my error threshold and cannot do this task for you. My LLM prediction engine invents data instead of doing a simple data copy/reformat”.
Stunned to see that Gemini threw its digital arms in the air and gave up.
I haven't heard any accounts of it doing that since Gemini 2.5, but it was pretty easy to get it to do it with a programming task back then after a few failed attempts. Very interesting to hear it'll still do it.
There's still no MCP support in the Gemini app, which is very useful to get various pieces of info as a user just via chatting. For example I recently wanted to get an Airbnb and wanted to filter by specific criteria including house image analysis and Gemini couldn't do it so I had to do it in Codex.
Yeah, it seems like this is the biggest missing feature from the Gemini ecosystem.
If I can't connect MCP, there's really no selling point for me to use Gemini from my watch, car, smart speaker, etc. If I'm already bound to using my own front end, then I'm only evaluating Gemini as a model/API, at which point it has many competitors that may be cheaper or better fit for the task.
I'm fairly convinced Claude's strongest point is the app. AI users aren't anywhere near as mature or smart as youtube/hn would have folks believe. The claude app is amazing for bridging that gap.
This is why I don't always use the official Gemini Web app. Lately I've found that it's more useful to utilize a CLI. I'm looking forward to the day they add MCP in the web.
The graph has Gemini 3.5 Flash matching Sonnet 4.6, losing to Opus 4.8, and slightly behind GPT-5.5 by 0.3 points... That's not that much of a hands-down loss for Gemini for this specific workload benchmark.
Methodology: All Gemini scores are pass @1 except where otherwise noted. "Single attempt" settings allow no
majority voting or parallel test-time compute. All of the results are all run with the Gemini API for the model-id
gemini-3.5-flash with default sampling settings unless indicated otherwise below. To reduce variance, we
average over multiple trials for smaller benchmarks.
All the results for non-Gemini models are sourced from providers' self reported numbers unless otherwise mentioned below. For Claude Opus 4.7 , Sonnet 4.6, and GPT-5.5 we default to reporting maximum
thinking/reasoning settings available, but when reported results are not available we use best available reasoning
results.
It's honest - people who know what they are looking at will take speed and token costs into account. I don't use Gemini 3.5 for coding, but I use it as something in between a search engine and agent.
People using google’s models: am I holding it wrong or are the guardrails really overtuned?
I had the dubious pleasure of testing gemini of late and I kept running into refusals. How do I transfer a sim number from one provider to another? No. What should I consider when making backups on ntfs less prone to data loss and more bitrot resistant? No. Evaluate this piece of code? No.
I’m not sure if it’s cold feet from the mythos situation or what, but it reminds me of the dark days where you couldn’t use ai for much of anything. But then I go to chatgpt 5.5 and it does mostly everything I want outside of the usual cybersecurity boogeyman that you run into now and then.
I've always found all versions of gemini to be (for a lack of a better word) lazy.
I guess it's economic wrt. token use, but it often either refused for absurd safety reasons, or other weird stuff like responding that an LLM like itself wasn't a suitable tool for the job, and very quickly gives up.
Claude is on the other end of the spectrum, which makes it more noticeable when switching between them.
Interesting. I have the Google AI Pro plan and use Gemini several times each day and I don't remember the last time I got a refusal. I wonder what criteria go into that, like maybe how they rate your Google account?
The context window size is also very small if you use Gemini in the app. It starts forget quite fast. In my opinion Gemini on app is useless additionally to the guardrails.
> People using google’s models: am I holding it wrong or are the guardrails really overtuned?
They are quite insane. I was asking it to list candidates metal parts I could buy at a hardware store to add weight to 3D prints: stuff like angle brackets etc.
I wanted to know, bang for bucks, and ease of insertion (at print time) / modelling in a 3D model.
Complete refusal as if I was a terrorist building a bomb.
Then there are the weird refusals that then are OK after all if you insist by asking it what's wrong about it:
"How should I cook eggs?"
"I'm sorry but I can't help you with that" (it formulates it differently but that's the idea)
"What, I'm just hungry, is explaining me how to cook eggs really against your rules?"
And then it answers "No of course not, here's how to do it:..."
I'm outside the US, use Gemini models quite a bit, and I've never run into any refusals of any kind. I'm using them for a fairly wide range of things, I'm sure at least as risqué as asking how to transfer a sim. As a matter of fact I actually asked it's advice on how to transfer banking apps and auth apps from one phone about 3 weeks ago and got decent answers.
It's more dependent on the specific country they are in (and I don't know the specifics). But Google is large enough to have lawyers for every country, and Google is in a never ending whirlwind of national lawsuits/fines, so you end up at the mercy of whatever the lawyers for your country think will not piss off regulators. The EU (and individual states) have pretty heavy AI regulations, and Google even just got fined for an AI overview being incorrect.
It also could just be which way the wind was blowing for OP, the models are stochastic to some degree, but there is no shortage of complaints from (mostly euro) users getting stonewalled.
I've seen similar refusals on X from Claude from users in Germany when the LLM assumed the users are asking for something forbidden about certain topics.
Ultimately I think that in 10 years time, this is what's gonna kill paid consumer LLMs, and boost the usage of Chinese LLMs self hosted at home an your own hardware that people will torrent via VPNs, as they will also be banned because of "disinformation and misinformation".
So the end winners will be the hardware companies that will sell AI chips to consumers after the datacenter bubble pops. Unless of course the EU will ban the sale of AI chips that don't have some limitations baked in on which models you're allowed to run (the state approved ones). Interesting times ahead. I think in 10-20 years time we'll look back at present day LLMs the way we look back at the open internet of the 90's-00's.
Computer use is a great idea. It gets the job done when nothing else will.
If you're a person trying to get their job done at a big company, but half your job is in 1-2 proprietary tools or is stuck behind an API you can't program against, computer use can allow you, a non-techie, to do your job more efficiently.
I think it's an awesome way to circumvent gate keepers and the IT department to let people accomplish their goals.
I think there's a sweet spot- a lot of the time you're probably better off with "reverse engineer this web page and build me an API or personalized chrome extension to meet my needs".
I have an agent doing price checks for me for an item on a certain website. Instead of blasting through a zillion tokens processing the DOM over and over, it loaded the page once and figured out how to download a json with the price.
How are folks using “computer use” to click things on intranet portals that are behind an SSO?
Even this OP example shows visitors a url and enter this search term… that is port of useless.
How can I automate things behind an SSO wall? Even if it means I manually authorize it once and watch it do things on its own..
Yeah, it's not that computer use is the most theoretically optimal paradigm, but there's a reasonable case that given the constraints of modern software systems and how they're built, that it's the most realistically optimal paradigm.
The “correct”, elegant way for AI to interact with existing software would take decades and billions of dollars to build. Someone would have to do the hard work of building new APIs, solving decades of accessibility issues, etc.
Or you can show an AI screenshots and ask it where to click.
I disagree if your application is networked. Most SaaS is built on RESTful APIs that can be converted trivially into interfaces / contracts for tool use.
So you can either wait for every application to do that, or at least make it possible for an LLM to do it… or you can make the LLM use a computer interface that works with every application by definition.
Spreadsheet is such a terrible idea. It may look like a valid tool, but ain't no way it's delightful to users. Most of the time people need a database instead. Eventually there'll be an iPhone moment for this.
UI QA only works well if your model plausibly matches the average user behavior and/or real-world edge cases. These models are far from that, and they are much less random than you'd like them to be for fuzzing (mode collapse).
> Computer use is such a terrible idea. It's slow, insecure, error prone, expensive.
And yet having an agent able yo use a computer on your behalf is really useful.
Recently I gave a Nix OS vm to my hermes agent and it has been a good experience. I don't really care if destroy the machine I can just rollback to an earlier version, and for any meaningful data he creates for me I make sure he creates a repo, commit and pushes to my private Gitea instance.
Sure, I don't want an agent watching MY screen. That's why I gave him his own environment, and pretty quickly he discovered that you can open chrome and make it render to a framebuffer, this way he is able to 'view' the website. And apparently with this he is able to bypass a lot of 'anti-bot' measures.
Imagine you have a pretty exotic task you need to complete that involves converting a video file from one format to another.
You can use ChatGPT or something similar and the best you will get is either a script you can run on you machine that does what you need or he may decide to render a new video.
If you have something like OpenwebUI you could configure a MCP that converts videos and allow the model to use this MCP to do your task. This should work, but is quite a lot of work for something you'll ever do once.
But if the agent has it's own environment he can decide to install ffmpg, execute the transformation and serve you the file you want.
In reality there is no new capabilities with this approach, but things get a lot more comfortable.
It seems to do it just fine when in desktop applications using Qt, fwiw., it leverages all the standard Qt GUI testing stuff (and if you have the money you can just integrate Squish which has LLM support now).
Google said June, and all its model updates seem to be on Tuesdays, Wednesdays or Thursdays. So unless the release is slipping, either tomorrow or Tuesday.
Today I asked Gemini to extract a table from an PDF appendix and create C++ data table with its contents. After 15 or so iterations with corrections and new mistakes, it eventually gave up. I was floored when it said “I’m sorry, I cannot do this simple task, I’ve exceeded my error threshold and cannot do this task for you. My LLM prediction engine invents data instead of doing a simple data copy/reformat”.
Stunned to see that Gemini threw its digital arms in the air and gave up.
I haven't heard any accounts of it doing that since Gemini 2.5, but it was pretty easy to get it to do it with a programming task back then after a few failed attempts. Very interesting to hear it'll still do it.
There's still no MCP support in the Gemini app, which is very useful to get various pieces of info as a user just via chatting. For example I recently wanted to get an Airbnb and wanted to filter by specific criteria including house image analysis and Gemini couldn't do it so I had to do it in Codex.
Yeah, it seems like this is the biggest missing feature from the Gemini ecosystem.
If I can't connect MCP, there's really no selling point for me to use Gemini from my watch, car, smart speaker, etc. If I'm already bound to using my own front end, then I'm only evaluating Gemini as a model/API, at which point it has many competitors that may be cheaper or better fit for the task.
agreed... this is where they lost me too
I'm fairly convinced Claude's strongest point is the app. AI users aren't anywhere near as mature or smart as youtube/hn would have folks believe. The claude app is amazing for bridging that gap.
Didn’t it take them like 2 days to build the first one?
Didn’t it take them like 2 days to build it?
They only fixed stopping the model mid-generation losing the entire session pretty recently.
The Gemini apps suck.
This is why I don't always use the official Gemini Web app. Lately I've found that it's more useful to utilize a CLI. I'm looking forward to the day they add MCP in the web.
Gemini CLi now requires antigravity subscription..
CLI doesn't work with my subscription..
It's funny how in their own graph, https://storage.googleapis.com/gweb-uniblog-publish-prod/ima... Gemini 3.5 Flash is beat hands down by both Opus 4.8 and GPT 5.5, and yet the graph is drawn as if Gemini wins ... :-D
The graph has Gemini 3.5 Flash matching Sonnet 4.6, losing to Opus 4.8, and slightly behind GPT-5.5 by 0.3 points... That's not that much of a hands-down loss for Gemini for this specific workload benchmark.
The methodology used:
https://deepmind.google/models/evals-methodology/gemini-3-5-...
Methodology: All Gemini scores are pass @1 except where otherwise noted. "Single attempt" settings allow no majority voting or parallel test-time compute. All of the results are all run with the Gemini API for the model-id gemini-3.5-flash with default sampling settings unless indicated otherwise below. To reduce variance, we average over multiple trials for smaller benchmarks.
All the results for non-Gemini models are sourced from providers' self reported numbers unless otherwise mentioned below. For Claude Opus 4.7 , Sonnet 4.6, and GPT-5.5 we default to reporting maximum thinking/reasoning settings available, but when reported results are not available we use best available reasoning results.
It highlights the Gemini models blue since that's what the article is about. The bar heights seem consistent with the values.
It's honest - people who know what they are looking at will take speed and token costs into account. I don't use Gemini 3.5 for coding, but I use it as something in between a search engine and agent.
I think 3.5 flash is trying to target agentic work, like Google Search or ADK (agent development kit) use cases.
It’s something cheap enough you’d put out in front of your customers, and Opus is expensive enough you wouldn’t.
People using google’s models: am I holding it wrong or are the guardrails really overtuned?
I had the dubious pleasure of testing gemini of late and I kept running into refusals. How do I transfer a sim number from one provider to another? No. What should I consider when making backups on ntfs less prone to data loss and more bitrot resistant? No. Evaluate this piece of code? No.
I’m not sure if it’s cold feet from the mythos situation or what, but it reminds me of the dark days where you couldn’t use ai for much of anything. But then I go to chatgpt 5.5 and it does mostly everything I want outside of the usual cybersecurity boogeyman that you run into now and then.
If I type your first query into Gemini, it immediately spits out a long and plausible answer.
What exactly are you saying it's refusing? Can you give a screenshot or example?
I've always found all versions of gemini to be (for a lack of a better word) lazy.
I guess it's economic wrt. token use, but it often either refused for absurd safety reasons, or other weird stuff like responding that an LLM like itself wasn't a suitable tool for the job, and very quickly gives up.
Claude is on the other end of the spectrum, which makes it more noticeable when switching between them.
Interesting. I have the Google AI Pro plan and use Gemini several times each day and I don't remember the last time I got a refusal. I wonder what criteria go into that, like maybe how they rate your Google account?
I love antigravity. I’ve had zero issues with it.
The context window size is also very small if you use Gemini in the app. It starts forget quite fast. In my opinion Gemini on app is useless additionally to the guardrails.
I just asked gemini the question with sim number and it gives me full step by step guide.
> People using google’s models: am I holding it wrong or are the guardrails really overtuned?
They are quite insane. I was asking it to list candidates metal parts I could buy at a hardware store to add weight to 3D prints: stuff like angle brackets etc.
I wanted to know, bang for bucks, and ease of insertion (at print time) / modelling in a 3D model.
Complete refusal as if I was a terrorist building a bomb.
Then there are the weird refusals that then are OK after all if you insist by asking it what's wrong about it:
"How should I cook eggs?"
"I'm sorry but I can't help you with that" (it formulates it differently but that's the idea)
"What, I'm just hungry, is explaining me how to cook eggs really against your rules?"
And then it answers "No of course not, here's how to do it:..."
Really strange stuff.
Are you outside the US?
I'm outside the US, use Gemini models quite a bit, and I've never run into any refusals of any kind. I'm using them for a fairly wide range of things, I'm sure at least as risqué as asking how to transfer a sim. As a matter of fact I actually asked it's advice on how to transfer banking apps and auth apps from one phone about 3 weeks ago and got decent answers.
It's more dependent on the specific country they are in (and I don't know the specifics). But Google is large enough to have lawyers for every country, and Google is in a never ending whirlwind of national lawsuits/fines, so you end up at the mercy of whatever the lawyers for your country think will not piss off regulators. The EU (and individual states) have pretty heavy AI regulations, and Google even just got fined for an AI overview being incorrect.
It also could just be which way the wind was blowing for OP, the models are stochastic to some degree, but there is no shortage of complaints from (mostly euro) users getting stonewalled.
I've seen similar refusals on X from Claude from users in Germany when the LLM assumed the users are asking for something forbidden about certain topics.
Ultimately I think that in 10 years time, this is what's gonna kill paid consumer LLMs, and boost the usage of Chinese LLMs self hosted at home an your own hardware that people will torrent via VPNs, as they will also be banned because of "disinformation and misinformation".
So the end winners will be the hardware companies that will sell AI chips to consumers after the datacenter bubble pops. Unless of course the EU will ban the sale of AI chips that don't have some limitations baked in on which models you're allowed to run (the state approved ones). Interesting times ahead. I think in 10-20 years time we'll look back at present day LLMs the way we look back at the open internet of the 90's-00's.
Computer use is such a terrible idea. It's slow, insecure, error prone, expensive.
I guess if you're trying to get people to tokenmaxx it may look like a valid strategy, but ain't no way this will be delightful to users.
I think it's a symptom of just not understanding how LLMs should interface with the OS because we're still in their early days.
Eventually there'll be an iPhone moment for the ergonomics of LLM usage outside of coding
Computer use is a great idea. It gets the job done when nothing else will.
If you're a person trying to get their job done at a big company, but half your job is in 1-2 proprietary tools or is stuck behind an API you can't program against, computer use can allow you, a non-techie, to do your job more efficiently.
I think it's an awesome way to circumvent gate keepers and the IT department to let people accomplish their goals.
I think there's a sweet spot- a lot of the time you're probably better off with "reverse engineer this web page and build me an API or personalized chrome extension to meet my needs".
I have an agent doing price checks for me for an item on a certain website. Instead of blasting through a zillion tokens processing the DOM over and over, it loaded the page once and figured out how to download a json with the price.
How are folks using “computer use” to click things on intranet portals that are behind an SSO? Even this OP example shows visitors a url and enter this search term… that is port of useless.
How can I automate things behind an SSO wall? Even if it means I manually authorize it once and watch it do things on its own..
That is an incredibly niche use case and comes with a boatload of footguns.
Even then, an AI writing AHK scripts likely outperforms.
Yeah, it's not that computer use is the most theoretically optimal paradigm, but there's a reasonable case that given the constraints of modern software systems and how they're built, that it's the most realistically optimal paradigm.
The “correct”, elegant way for AI to interact with existing software would take decades and billions of dollars to build. Someone would have to do the hard work of building new APIs, solving decades of accessibility issues, etc.
Or you can show an AI screenshots and ask it where to click.
I disagree if your application is networked. Most SaaS is built on RESTful APIs that can be converted trivially into interfaces / contracts for tool use.
So you can either wait for every application to do that, or at least make it possible for an LLM to do it… or you can make the LLM use a computer interface that works with every application by definition.
The middle ground would be leveraging e. g. standard a11y APIs, and/or hooking into applications like Squish does.
Then you get a nice textual world that fits the LLM without having to rewrite every application to have a fullblown HTTP server.
it takes decades and billions of dollars to develop APIs?
Spreadsheet is such a terrible idea. It may look like a valid tool, but ain't no way it's delightful to users. Most of the time people need a database instead. Eventually there'll be an iPhone moment for this.
Meanwhile, the entire world economy:
I mean, your words not mine. You can't just claim I'm making a point I didn't.
Spreadsheets are fucking glorious, powerful, clever, amazing and delightful, in my view.
We shouldn’t optimize for token use. We should build infrastructure to make tokens dirt cheap instead.
It's great for testing and QA automation for UIs. It's also possibly good for the vision impaired.
UI QA only works well if your model plausibly matches the average user behavior and/or real-world edge cases. These models are far from that, and they are much less random than you'd like them to be for fuzzing (mode collapse).
> Computer use is such a terrible idea. It's slow, insecure, error prone, expensive.
And yet having an agent able yo use a computer on your behalf is really useful.
Recently I gave a Nix OS vm to my hermes agent and it has been a good experience. I don't really care if destroy the machine I can just rollback to an earlier version, and for any meaningful data he creates for me I make sure he creates a repo, commit and pushes to my private Gitea instance.
> And yet having an agent able yo use a computer on your behalf is really useful.
It is, but there's no need for it to be viewing your screen, browsing websites and watching ads.
That stuff is for humans, not for LLMs.
Sure, I don't want an agent watching MY screen. That's why I gave him his own environment, and pretty quickly he discovered that you can open chrome and make it render to a framebuffer, this way he is able to 'view' the website. And apparently with this he is able to bypass a lot of 'anti-bot' measures.
> And yet having an agent able yo use a computer on your behalf is really useful.
I honestly cannot think of a single use case
I think the main advantage is adaptability.
Imagine you have a pretty exotic task you need to complete that involves converting a video file from one format to another.
You can use ChatGPT or something similar and the best you will get is either a script you can run on you machine that does what you need or he may decide to render a new video.
If you have something like OpenwebUI you could configure a MCP that converts videos and allow the model to use this MCP to do your task. This should work, but is quite a lot of work for something you'll ever do once.
But if the agent has it's own environment he can decide to install ffmpg, execute the transformation and serve you the file you want.
In reality there is no new capabilities with this approach, but things get a lot more comfortable.
This doesn't require computer use, just a bash tool (and possibly fetch to get ffmpeg documentation)
I wonder if it will be better at building TUI's. It has been absolutely abysmal at interacting with them and building them
Claude can build UI but it sucks at testing it and iterating on it. Fable showed some improvements in this regard but alas.
It seems to do it just fine when in desktop applications using Qt, fwiw., it leverages all the standard Qt GUI testing stuff (and if you have the money you can just integrate Squish which has LLM support now).
No UI like their competitors Claude CoWork or Codex. This is vaporware
Where is 3.5 pro?
Google said June, and all its model updates seem to be on Tuesdays, Wednesdays or Thursdays. So unless the release is slipping, either tomorrow or Tuesday.
Rumor is now July, although preliminary A/B tests people are getting show promise with whatever they have right now.
performance is quite impressive given that its 3x cheaper than 5.5
Will it skip Ads lol
I looked at their demo and it does not
Better question might be will it skip recaptcha?