I don't find this very compelling. If you look at the actual graph they are referencing but never showing [1] there is a clear improvement from Sonnet 3.7 -> Opus 4.0 -> Sonnet 4.5. This is just hidden in their graph because they are only looking at the number of PRs that are mergable with no human feedback whatsoever (a high standard even for humans).
And even if we were to agree that that's a reasonable standard, GPT 5 shouldn't be included. There is only one datapoint for all OpenAI models. That data point more indicative of the performance of OpenAI models (and the harness used) than of any progression. Once you exclude it it matches what you would expect from a logistic model. Improvements have slowed down, but not stopped
If you measure completion rate on a task where a single mistake can cause a failure, you won't see noticeable improvements on that metric until all potential sources of error are close to being eliminated, and then if they do get eliminated it causes a sudden large jump in performance.
That's fine if you just want to know whether the current state is good enough on your task of choice, but if you also want to predict future performance, you need to break it down into smaller components and track each of them individually.
> until all potential sources of error are close to being eliminated
This is what PSP/TSP did - one has to (continually) review its' own work to identify most frequent sources of (user facing) defects.
> if you also want to predict future performance, you need to break it down into smaller components and track each of them individually.
This is also one of tenets of PSP/TSP. If you have a task with estimate longer that a day (8 hours), break it down.
This is fascinating. LLM community discovers PSP/TSP rules that were laid over more than twenty years ago.
What LLM community miss is that in PSP/TSP it is an individual software developer who is responsible to figure out what they need to look after.
What I see is that it is LLM users who try to harness LLMs with what they perceive as errors. It's not that LLMs are learning, it is that users of LLMs are trying to stronghold these LLMs with prompts.
I don't know that graph to me shows Sonnet 4.5 as worse than 3.7. Maybe the automated grader is finding code breakages in 3.7 and not breaking that out? But I'd much prefer to add code that is a different style to my codebase than code that breaks other code. But even ignoring that the pass rate is almost identical between the two models.
1) Something happened during 2025 that made the models (or crucially, the wrapping terminal-based apps like Claude Code or Codex) much better. I only type in the terminal anymore.
2) The quality of the code is still quite often terrible. Quadruple-nested control flow abounds. Software architecture in rather small scopes is unsound. People say AI is “good at front end” but I see the worst kind of atrocities there (a few days ago Codex 5.3 tried to inject a massive HTML element with a CSS before hack, rather than proprerly refactoring markup)
Two forces feel true simultaneously but in permanent tension. I still cannot make out my mind and see the synthesis in the dialectic, where this is truly going, if we’re meaningfully moving forward or mostly moving in circles.
I only say that because I'm a shit frontend dev. Honestly, I'm not that bad anymore, but I'm still shit, and the AI will probably generate better code than I will.
As long as humans are needed to review code, it sounds your role evolves toward prompting and reviewing.
Which is akin to driving a car - the motor vehicle itself doesn’t know where to go. It requires you to prompt via steering and braking etc, and then to review what is happening in response.
That’s not necessarily a bad thing - reviewing code ultimately matters most. As long as what is produced is more often than not correct and legible.. now this is a different issue for which there isn’t a consensus across software engineer’s.
There is a decent case for this thesis to hold true especially if we look at the shift in training regimes and benchmarking over the last 1-2 years. Frontier labs don't seem to really push pure size/capability anymore, it's an all in focus on agentic AI which is mainly complex post-training regimes.
There are good reasons why they don't or can't do simple param upscaling anymore, but still, it makes me bearish on AGI since it's a slow, but massive shift in goal setting.
In practice this still doesn't mean 50 % of white collar can't be automated though.
> In practice this still doesn't mean 50 % of white collar can't be automated though.
Let me ask you this, though: if we wanted to, what percentage of white collar jobs could have been automated or eliminated prior to LLMs?
Meta has nearly 80k employees to basically run two websites and three mobile apps. There were 18k people working at LinkedIn! Many big tech companies are massive job programs with some product on the side. Administrative business partners, program managers, tech writers, "stewards", "champions", "advocates", 10-layer-deep reporting chains... engineers writing cafe menu apps and pet programming languages... a team working on in-house typefaces... the list goes on.
I can see AI producing shifts in the industry by reducing demand for meaningful work, but I doubt the outcome here is mass unemployment. There's an endless supply of bs jobs as long as the money is flowing.
Meta has 80k employees to run the world's most massive engine of commerce through advertising and matching consumers to products.
They build generative AI tools so people can make ads more easily.
They have some of the most sophisticated tracking out there. They have shadow profiles on nearly everyone. Have you visited a website? You have a shadow profile even if you don't have a Facebook account. They know who your friends are based on who you are near. They know what stores you visit when.
Large fractions of their staff are making imperceptible changes to ads tracking and feed ranking that are making billions of dollars of marginal revenue.
What draws you in as a consumer is a tiny tip of the iceberg of what they actually do.
There are many reasons why we are seeing cuts economically, but the fact that it is possible to make such large cuts is because there were way too many people working at these companies. They had so much cheap money that they over-hired, now money isn't so cheap and they need to reduce headcount. AI need not enter the conversation to get to that point.
This is unfair and dismissive of many roles. Coordination in a massive, technically complex company that has to adhere to laws and regulations is a critical role. I don't get why people shit on certain roles (I'm a SWE). Our PgMs reduce friction and help us be more productive and focused. Technical writers produce customer-facing content and code, and have nothing to do with supporting internal bureaucracy. There are arguments against this in Bullshit Jobs but do you think companies pay PgMs or HR employees hundreds of thousands of dollars a year out of the goodness of their own hearts? Or maybe they actually help the business?
You realize that the reason you need to manage this organizational complexity is largely because the organization is so huge?...
The reality is that you could run LinkedIn with far, far fewer people. You probably need fewer than 100 for core engineering, and likely less than 1,000 overall if you include compliance, sales, and so on - especially since a lot of overseas compliance stuff is outsourced to consulting firms, it's not like you have a team of lawyers in every country in the world.
Before there was so much money in the system, we used to run companies that way. Two decades ago, I worked for a company that had tens of millions of users, maintained its own complex nationwide infra (no AWS back then), and had 400 full-time employees. That made coordination problems a lot easier too. We didn't need ten layers of people and project management because there just wasn't that many of us.
When doubling the number of employees can triple your revenue, you do it.
Keeping a website running with high uptime is not the goal. Maximizing revenue and profit is. The extra people aren't waste, they're what drive the incremental imperceptible changes that make these companies profitable.
fwiw the merge rate metric itself might be misleading. most real codebases have implicit conventions and architectural patterns that aren't captured in the issue description, so even if the model writes correct code it might not match what the maintainer actually wanted. imo the bigger signal is how much back-and-forth it takes before merging, not whether the first attempt lands cleanly.
I am pretty convinced that for most types of day to day work, any perceived improvements from the latest Claude models for example were total placebo. In blind tests and with normal tasks, people would probably have no idea if they're using Opus 4.5 or 4.6.
This has basically been my experience since Sonnet 3.5. I've been working on a personal project on and off with various models and things since then and the biggest difference between then and now is that it will do larger chunks of work than it did before, but the quality of the code is not particularly better, I still have to do a lot of cleanup and it still goes off the rails pretty frequently. I have to do fewer individual prompts, but the time spent reviewing the code takes longer because I also have to mentally process and fix larger chunks of code too
Is it a better user experience now? Yes. Has it boosted my productivity on this project? Absolutely.
But it still needs a ton of hand holding for anything complicated and I still deal with tons of "OK, this bug is fixed now!" followed by manually confirming a bug still exists.
Very possible. Some things work like a charm on first try for me, others you can spell it out again and again. And then yet again. Something to do with training data, obviously.
Interesting article, although with so few data points and such a specific time slice it is difficult to draw serious conclusions about the "improvement" of LLM models.
It's notably lacking newer models (4.5 Opus, 4.6 Sonnet) and models from Gemini.
LLMs appear to naturally progress in short leaps followed by longer plateaus, as breakthroughs are developed such as chain-of-thought, mixture-of-experts, sub-agents, etc.
Focusing on flashy breakthroughs hides the issue that bigger models and merge benchmarks rarely translate to reliability in real codebases. For routine merges, subtle regressions and context quirks matter more than headline progress. Unless evals stress nasty scenarios like multi-file renames with tricky conflicts, the numbers are mostly for show. Progress will plateau until someone tunes for the boring, messy cases that waste dev time.
I don't think it's true, but am I alone in wishing it was? My world is disrupted somewhat but so far I don't think we have a thing that upends our way of life completely yet. If it stayed exactly this good I'd be pretty content.
I agree with your sentiment, but I think we've yet to see the full application of the current technology. (Even if LLMs themselves don't improve, there's significant opportunity for people to use it in ways not currently being done)
When the dust settles, for example if LLM's were to stop improving today, we would come to learn their exact capabilities, what they can do reliably and what they can't.
Once we know what they can do well and how to get them to do it well, and what they can't, you could say we "trust" them to do the first category well and just stop trying to get it to do the second category.
That's a big problem with very specific manifestations. My startup helps customers handle regulatory compliance, also by forwarding complex questions to a pool of consultants.
We've compared now more than a hundred replies to that of GPT Pro, and the quality is roughly the same. Sometimes a little worse, sometimes a little better. Always more detailed. Never unacceptable.
But how to convince our customers that we have the right technology and know how to use it appropriately? We're trying, but it's not easy.
Part of that's accountability. In the event of the LLM producing rubbish, as rare as it may be, who is accountable? There is not a person and her reputation attached to it.
LLM plus human should be better than either standalone. You won’t be able to make as much money scaling out, though.
You won’t be able to scale out and make as much money though. But surely you’re not only concerned about profit, right? What’s the point of life if you’re just trying to get rich.
That's an interesting claim, but I don't see it in my own work. They have got better but it's very hard to quantify. I just find myself editing their work much less these days (currently using GPT 5.4).
Without meaning to sound dismissive, because I'm really not intending to, there's also the possibility that you've gotten worse after enough time using them. You're treating yourself as a constant in this, but man cannot walk in the same river twice.
This is such a silly response when "You've gotten better at using them and know how to work around their flaws now." is right there and seems a lot more plausible.
That's a possibility, but I doubt it. I've been programming for 35 years and know what I like in code. I've also previously maintained a long review prompt in which I tell the models all the ways in which they get things wrong and to go look for/fix those problems. But those review passes now don't take as long because there are fewer such problems to begin with.
In particular GPT 5.4 is much better at not duplicating code unnecessarily. It'll take the time to refactor, to search for pre-existing utility functions, etc.
The problem with evals is the underlying rubric will always be either subjective, or a quantitative score based on something that is likely now baked into the training set directly.
You kind of have to go on "feels" for a lot of this.
Controversial opinion from a casual user, but state-of-art LLMs now feel to me more intelligent then the average person on the steet. Also explains why training on more average-quality data (if there's any left) is not making improvements.
But LLMs are hamstrung by their harnesses. They are doing the equivalent of providing technical support via phone call: little to no context, and limited to a bidirectional stream of words (tokens). The best agent harnesses have the equivalent of vision-impairment accessibility interfaces, and even those are still subpar.
Heck, giving LLMs time to think was once a groundbreaking idea. Yesterday I saw Claude Code editing a file using shell redirects! It's barbaric.
I expect future improvements to come from harness improvements, especially around sub agents/context rollbacks (to work around the non-linear cost of context) and LLM-aligned "accessibility tools". That, or more synthetic training data.
entirely so. i think anthropic updated something about the compact algorithm recently, and its gone from working well over long times to basically garbage whenever a compact happens
And search engines are narrow tools that can only output copies of its dataset. An LLM is capable of surprisingly novel output, even if the exact level of creativity is heavily debated.
I feel even if the models are stagnating, the tooling around them, and the integrations and harnesses they have are getting significantly more capable (if not always 'better' - the recent vscode update really handicapped them for some reason). Things like the new agent from booking.com or whatever, if it could integrate with all hotels, activities, mapping tools, flight system, etc could be hugely powerful.
Assuming we get no better than opus 4.6, they're very capable. Even if they make up nonsense 5% of the time!
I think what happened with static image generation is happening with LLMs. Basically the tools around are becoming better, but all the AI improvements stall, the error rate stay the same (but external tools curate the results so it won't be noticeable if you don't run your own model), the accuracy is still slightly improving, but slower and slower, and never reach the 'perfect' point. Basically stablediffusion early 2025
Image quality has improved a lot in recent months thanks to better models. The ability of people to notice these improvements is plateauing because they are not trained to spot artifacts, which are becoming more obscure.
Yes, slight increase in that kind of accuracy. And newer models still generate absurd stuff. Ask for an historical picture, like 'a London market in the 18th century', and it is still as historically wrong as it was 2 years ago. It is useful for fantasy/sci-fi though, I use them a lot. But I don't see the point of newer models since late 2024.
I have no idea how a London market in the 18th century should look like so I cannot challenge that but recent models like nano banana ones, Z-image (on a lesser extend) can generate images that are essentially indistinguishable from actual stock photos, this wasn't true for late 2024 models, with wonky backgrounds, too smooth skins and general lack of details (the classic AI-look that AI images had).
Mmm, I mostly generate scenes, not people, and when I do, it's always in a not photorealistic way. It is true that newer models have more details, I think that I put that into 'slight accuracy improvements', is it really major? Or is it mostly for close up people/animals and it is likely that I just didn't notice?
It will be wrong on a lot of details. Basically you would get a market scene that feels 18th century gb, but will use 18th century russian/french/Austrian details, or 19th century/20th century British artefacts, or a mix of both. And the further you go from western places, the higher the error rate is. Basically generating fiction scenes. That's pretty much my usecase, so that's fine, but I won't ever use it to illustrate a historical TTrpg.
Truth is I'm probably wrong. I should keep on testing ... but at the same time I precisely gave up because I didn't think the trend was fast enough to keep on investing on checking it so frequently. Now I just read this kind of post, ask around (mainly arguing with comments asking for genuine examples that should be "surprising" and kept on being disappointed) and that seems to be enough for a proxy.
I should though, as I mentioned in another comment, keep track of failed attempts.
PS: I check solely on self-hosted models (even if not on my machine but least on machines I could setup) because I do NOT trust the scaffolding around proprietary closed sources models. I can't verify that nobody is in the loop.
Benchmaxxing aside, if you are using those tools for programming on a regular basis it should be self-evident that they are improving. I find it very hard to believe that someone using LLMs today vs what was available one year ago (Claude Code released Feb 2025) would have any difficulty answering this question.
I think it is important to try to find more rigorous things to test than the general sentiment of the people using the tools. If only because the more benchmarks we have the more we can improve models without regressions. METR is asking a really interesting question here, "are models improving at making one shot PRs?". The answer seems to be, yes, but slower than benchmarks suggest, if you look at the pass rate of different versions of Claude Sonnet. A reasonable answer is "you're not supposed to use them by making one shot PRs", but then ideally we would need to have some kind of standarized test for the ability of models to incorporate feedback and evolve PRs.
Well, on one hand they lack new data. Lot's of new code came out of an LLM, so it feeds back.
On the other hand, LLMs tend to go for an average by their nature (if you squint enough). What's more common in their training data, it's more common in the output, so getting them better without fundamental changes, requires one to improve the training data on average too which is hard.
What did improve a lot is the tooling around them. That's gotten way better.
> Well, on one hand they lack new data. Lot's of new code came out of an LLM, so it feeds back.
Supposedly model curation is a Big Deal at Big AI, and they're especially concerned about Ouroboros effects and poisoned data. Also people are still contributing to open source and open sourcing new projects, something that should have slowed to trickle by 2023, once it became clear that from now on, you're just providing the fuel for the machines that will ultimately render you unemployable (or less employable), and that these machines will completely disregard your license terms, including those of the most permissive licenses that seek only attribution, and that you're doing all of this for free.
You really can't model these 5 data points with a linear regression or a step function. The models are of different sizes / use cases, and from two different labs. I feel like what we've observed generally is that different labs releasing similarly sized models at similar times are generally pretty similar.
I think the only reasonable thing to read into is Sonnet 3.5 -> 3.7 -> 4.5. But yeah, you just can't draw a line through this thing.
I will die on the hill that LLMs are getting better, particularly Anthropic's releases since December. But I can't point at a graph to prove that, I'm just drawing on my personal experience. I do use Claude Code though, so I think a large part of the improvement comes from the harness.
These studies are always really hard to judge the efficacy of. I would say though the most surprising thing to me about LLMs in the past year is how many people got hyped about the Opus 4.5 release. Having used Claude Code at work since it was released I haven't really noticed any step changes in improvement. Maybe that's because I've never tried to use it to one shot things?
Regardless I'm more inclined to believe that 4.5 was the point that people started using it after having given up on copy/pasting output in 2024. If you're going from chat to agentic level of interaction it's going to feel like a leap.
I used it with Sonnet 4.0 a lot, and there was vastly more back-and-forth and correction of "dumb" things, such as forgetting to add "using" statements in C# files.
I don't know if it's model, or harness improvements, or inbuilt-memory or all of the above, but it often has a step where it'll check itself that is done now before trying to build and getting an inevitable failure.
Those small things add up to a much smoother and richer experience today compared to 6 months ago.
My experience has been that raw “one-shot intelligence” hasn’t improved as dramatically in the last year, but the workflow around the models has improved massively.
I've been able to supercharge a hobby project of mine over the last couple months using Opus 4.6 in claude code. I had to collaborate and write code still, but claude did like 75% of the work to add meaningful new features to an iOS/Android native mobile app, including Live Activities which is so overly complicated i would not have been able to figure that out. I have it running in a folder that contains both my back end api (express) and my mobile app (nativescript), so it does back end and front end work simultaneously to support new features. this wasnt possible 8 months ago.
I feel like anyone used AI coding tools before 11/25 and after 1/26 (with frontier models) will say there has been a massive jump in, there is a difference between whether LLM can do a specific task or pass some arguably arbitrary checks by maintainers vs. what the are capable of.
We still have tons of gaps about how to build and maintain code with AI, but LLM themselves getting better at an unbelievable pace, even with this kind of data analysis I’m surprised anyone can even question it.
> I feel like anyone used AI coding tools before 11/25 and after 1/26 (with frontier models) will say there has been a massive jump in, there is a difference between whether LLM can do a specific task or pass some arguably arbitrary checks by maintainers vs. what the are capable of.
How much of that is the model and how much of that is the tooling built around it? Also why is the tooling, specifically Claude Code, so buggy?
It's my experience that opus 4, and then, particularly, 4.5, in Claude code, are head and shoulders above the competition.
I wrote an agentic coder years ago and it yielded trash. (Tried to make it do then what kiro does today).
The models are better. Now, caveat - I don't use anything but opus for coding - Sonnet doesn't do the trick. My experience with Codex and Gemini is that their top models are as good as Sonnet for coding...
I was trying to do something yestesrday and Claude was keep messing it up, after like an hour i realized the model somehow switched to sonet, opus 4.6 is crazy good. It’s very obvious in practice.
Although I feel like for chasing bugs and big systems codex is even better
This has been the general consensus for about three years now. "Drastic increases in capability have happened the last 3-6 months" have been a constant refrain.
Without any data from the study past September I think its not unreasonable, if you want to make an argument based on evidence.
For me personally, I agree with you, I'm really seeing it as well.
> "Drastic increases in capability have happened the last 3-6 months" have been a constant refrain.
well, yeah. because that's been the experience for many people.
3 years ago, trying to use ChatGPT 3.5 for coding tasks was more of a gimmick than anything else, and was basically useless for helping me with my job.
today, agentic Opus 4.6 provides more value to me than probably 2 more human engineers on my team would
I tried GPT3.5 for translating code from typescript to rust. It made many mistakes in rust. It couldn't fix borrow checker issues. The context was so small that I could only feed it small amounts of my program at a time. It also introduced new bugs into the algorithm.
Yesterday I had an idea for a simple macos app I wanted. I prompted claude code. It programmed the whole thing start to finish in 10 minutes, no problem. I asked it to optimize the program using a technique I came up with, and it did. I asked it to make a web version and it did. (Though for some reason, the web version needed several rounds of "it doesn't work, here's the console output").
I'm slowly coming to terms with the idea that my job is fundamentally changing. I can get way more done by prompting claude than I can by writing the code myself.
There's a consensus that SOMETHING changed with Opus 4.5. It might have been the "merge rates" metric, it might have not.
I'm certainly getting faster and cleaner-looking solutions for certain issues on Opus 4.6 than I was 5 months ago, but I'm not sure about the ability to solve (or even weigh in) the actual hard stuff, i.e. the stuff I'm paid for.
And I'm definitely not sure about the supposed big step between 4.5 and 4.6. I'm literally not seeing any.
I agree completely. I haven't noticed much improvement in coding ability in the last year. I'm using frontier models.
What's been the game changer are tools like Claude Code. Automatic agentic tool loops purpose built for coding. This is what I have seen as the impetus for mainstream adoption rather than noticeable improvements in ability.
I write a lot of C++ and QML code. Codex 5.3, only released in Feb, is the the first model I've used that would regularly generate code that passes my 25 years expert smell test and has turned generative coding from a timesap/nuisance into a tool I can somewhat rely on not to set me back.
Claude still wasn't quite there at the time, but I haven't tried 4.6 yet.
QML is a declarative-first markup language that is a superset of the JavaScript syntax. It's niche and doesn't have a giant amount of training data in the corpus. Codex 5.3 is the first model that doesn't super botch it or prefers to write reams of procedural JS embeds (yes, after steering). Much reduced is also the tendency to go overboard on spamming everything with clouds of helper functions/methods in both C++ and QML. It knows when to stop, so to speak, and is either trained or able to reason toward a more idiomatic ideal, with far less explicit instruction / AGENTS.md wrangling.
It's a huge difference. It might be the result of very specific optimization, or perhaps simultaneous advancements in the harness play a bigger role, but in my books my kneck of the woods (or place on the long tail) only really came online in 2026 as far as LLMs are concerned.
Maybe n=1, but I disagree? I notice that Sonnet 4.6 follows instructions much better than 4.5 and it generates code much closer to our already in-place production code.
It's just a point release and it isn't a significant upgrade in terms of features or capabilities, but it works... better for me.
Are you using a tool like Claude Code or Codex or windsurf? I ask because I've found their ability to pull in relevant context improves tasks in exactly the way you're describing.
My own experience is that some things get better and some things get worse in perceived quality at the micro-level on each point release. i.e. 4.5->4.6
From my personal experience, they have gotten better, but they haven’t unlocked any new capabilities. They’ve just improved at what I was already using them for.
At the end of the day they still produce code that I need to manually review and fully understand before merging. Usually with a session of back-and-forth prompting or manual edits by me.
That was true 2 years ago, and it’s true now (except 2 years ago I was copy/pasting from the browser chat window and we have some nicer IDE integration now).
Yeah I'm not buying the last bit about lower MSE with one term in the model vs two (Brier with one outcome category is MSE of the probabilities). That's the sort of thing that would make me go dig to find where I fucked up the calculation.
With one term it gets more robust in the face of excluding endpoints when constructing the jackknife train/test split, I think. But you're right, it does sound fishy.
What the post is describing is just ANOVA. If removing a category improves the overall fit then fitting the two terms independently has the same optimal solution (with the two independent terms found to be identical). MSE never increases when adding a category.
This is why you have to reach to things that penalize adding parameters to models when running model comparisons.
Yesterday I asked a frontier model to help generate a report. It said great, it can do that, and output a table. I asked it to evaluate its prompt compliance in the result. It concluded that it had failed on every requirement. I asked why it had expressed such confidence, was it analagous to narcissism or psycopathy? It said no, and then said that if I just had to anthropomorphize it, I should think of it as a brilliant friend with severe frontal lobe brain damage.
Even if one-shot LLM performance has plateaued (which I'm not convinced this data shows given omission of recent models that are widely claimed to be better) that missing the point that I see in my own work. The improved tooling and agent-based approaches that I'm using now make the LLM one-shot performance only a small part of the puzzle in terms of how AI tools have accelerated the time from idea to decent code. For instance the planning dialogs I now have with Claude are an important part of what's speeding things up for me. Also, the iterative use of AI to identify, track, and take care of small coding tasks (none of which are particularly challenging in terms of benchmarks) is simply more effective. Could this all have been done with the LLM engines of late 2024. Perhaps, but I think the fine-tuning (and conceivably the system prompts) that make the current LLM's more effective at agent-centered workflows (including tool-use) are a big part of it. One-shot task performance at challenging tasks is an interesting, certainly foundational, metric. But I don't think it captures the important advances I see in how LLM's have gotten better over the last year in ways that actually matter to me. I rarely have a well-defined programming challenge and the obligation to solve it in a single-shot.
>This means the step function has more predictive power (“fits better”) than the linear slope. For fun, we can also fit a function that is completely constant across the entire timespan. That happens to get the best Brier score.
I mean, sure. but it's obvious in that graph that the single openai model is dragging down the right side. Wouldn't it be better to just stick to analyzing models from only one lab so that this was showing change over time rather than differences between models?
> This means llms have not improved in their programming abilities for over a year. Isn’t that wild? Why is nobody talking about this?
Because it's not true. They have improved tremendously in the last year, but it looks like they've hit a wall in the last 3 months. Still seeing some improvements but mostly in skills and token use optimization.
> but mostly in skills and token use optimization.
I have heard rumors that token use optimization has been a recent focus to try to tidy up the financials of these companies before they IPO. take that with a grain of salt though
LLM's have 100% gotten better, but it's hard to say if it's "intrinsically better", if that makes sense.
> OpenAI’s leading researchers have not completed a successful full-scale pre-training run that was broadly deployed for a new frontier model since GPT-4o in May 2024 [1]
That's evidence against "intrinsically better". They've also trained on the entire internet - we only have 1 internet, so.
However, late 2024 was the introduction of o1 and early 2025 was Deepseek R1 and o3. These were definitely significant reasoning models - the introduction of test time compute and significant RL pipelines were here.
Mid 2025 was when they really started getting integrated with tool calling.
Late 2025 is when they really started to become agentic and integrate with the CLI pretty well (at least for me). For example, codex would at least try and run some smoke tests for itself to test its code.
In early 2026, the trend now appears to be harness engineering - as opposed to "context engineering" in 2025, where we had to preciously babysit 1 model's context, we make it both easier to rebuild context (classic CS trick btw: rebooting is easier than restoring stale state [2]) and really lean into raw cli tool calling, subagents, etc.
FWIW, AI programming has still been as frustrating as it was when it was just TTC in 2025. Maybe because I don't have the "full harness" but it still has programming styles embedded such as silent fallback values, overly defensive programming, etc. which are obvoiusly gleaned from the desire to just pass all tests, rather than truly good programming design. I've been able to do more, but I have to review more slop... also the agents are really unpleasant to work with, if you're trying to have any reasonable conversation with them and not just delegate to them. It's as if they think the entire world revolves around them, and all information from the operator is BS, if you try and open a proper 2-way channel.
It seems like 2026 will go full zoom with AI tooling because the goal is to replace devs, but hopefully AI agents become actually nice to work with. Not sycophantic, but not passively aggressively arrogant either.
>To study how agent success on benchmark tasks relates to real-world usefulness, we had 4 active maintainers from 3 SWE-bench Verified repositories review 296 AI-generated pull requests (PRs). We had maintainers (hypothetically) accept or request changes for patches as well as provide the core reason they were requesting changes: core functionality failure, patch breaks other code or code quality issues.
I would also advise taking a look at the rejection reasons for the PRs. For example, Figure 5 shows two rejections for "code quality" because of (and I quote) "looks like a useless AI slop comment." This is something models still do, but that is also very easily fixable. I think in that case the issue is that the level of comment wanted hasn't been properly formalized in the repo and the model hasn't been able to deduce it from the context it had.
As for the article, I think mixing all models together doesn't make sense. For example, maybe a slope describe the increasing Claude Sonnet better than a step function.
Anecdotally, I haven't seen any real improvement from the AI tools I leverage. They're all good-ish at what they do, but all still lie occasionally, and all need babysitting.
I also wonder how much of the jump in early 2025 comes from cultural acceptance by devs, rather than an improvement in the tools themselves.
I think I'm coming to the same conclusion Gpt-3 to 5.3 have had real tangible but incremental improvements with quite diminishing returns.
Perhaps we won't see a phase change like improvement as we did from gpt-2 through to 3 until there is several more orders of magnitude parameters and/or training. Perhaps we will never see it again!
What is getting rapidly better is scaffolding but this seems to be more about understanding and building tools around LLMs than the LLMs themselves improving.
I'm still excited about AI but not constantly hyped to the rafters as some.
I think it depends on what you're using it for. If it is a simple kubernetes config then the model doesn't matter too much. Contract that with writing the scenario for a backtest for an algo that trades on a venue: it is not the same complexity and the basic models are terrible. I've had it tell me that it has added tests to find that they're just stubs!
Opus seems to be getting there, but on more complex tasks the others are a complete waste of time.
> If it is a simple kubernetes config then the model doesn't matter too much
I guess at least this person https://www.tomshardware.com/tech-industry/artificial-intell... might disagree. I think already to know what Kubernetes even is requires quite a bit of knowledge. Using a tool that manipulate its configuration files IN PRODUCTION without risking data loss is another ball game entirely.
And even if we were to agree that that's a reasonable standard, GPT 5 shouldn't be included. There is only one datapoint for all OpenAI models. That data point more indicative of the performance of OpenAI models (and the harness used) than of any progression. Once you exclude it it matches what you would expect from a logistic model. Improvements have slowed down, but not stopped
1: https://metr.org/assets/images/many-swe-bench-passing-prs-wo...
If you measure completion rate on a task where a single mistake can cause a failure, you won't see noticeable improvements on that metric until all potential sources of error are close to being eliminated, and then if they do get eliminated it causes a sudden large jump in performance.
That's fine if you just want to know whether the current state is good enough on your task of choice, but if you also want to predict future performance, you need to break it down into smaller components and track each of them individually.
This is fascinating. LLM community discovers PSP/TSP rules that were laid over more than twenty years ago.
What LLM community miss is that in PSP/TSP it is an individual software developer who is responsible to figure out what they need to look after.
What I see is that it is LLM users who try to harness LLMs with what they perceive as errors. It's not that LLMs are learning, it is that users of LLMs are trying to stronghold these LLMs with prompts.
It's useless and never gets better until it suddenly, unexpecty got good enough.
1) Something happened during 2025 that made the models (or crucially, the wrapping terminal-based apps like Claude Code or Codex) much better. I only type in the terminal anymore.
2) The quality of the code is still quite often terrible. Quadruple-nested control flow abounds. Software architecture in rather small scopes is unsound. People say AI is “good at front end” but I see the worst kind of atrocities there (a few days ago Codex 5.3 tried to inject a massive HTML element with a CSS before hack, rather than proprerly refactoring markup)
Two forces feel true simultaneously but in permanent tension. I still cannot make out my mind and see the synthesis in the dialectic, where this is truly going, if we’re meaningfully moving forward or mostly moving in circles.
I only say that because I'm a shit frontend dev. Honestly, I'm not that bad anymore, but I'm still shit, and the AI will probably generate better code than I will.
Which is akin to driving a car - the motor vehicle itself doesn’t know where to go. It requires you to prompt via steering and braking etc, and then to review what is happening in response.
That’s not necessarily a bad thing - reviewing code ultimately matters most. As long as what is produced is more often than not correct and legible.. now this is a different issue for which there isn’t a consensus across software engineer’s.
There are good reasons why they don't or can't do simple param upscaling anymore, but still, it makes me bearish on AGI since it's a slow, but massive shift in goal setting.
In practice this still doesn't mean 50 % of white collar can't be automated though.
Let me ask you this, though: if we wanted to, what percentage of white collar jobs could have been automated or eliminated prior to LLMs?
Meta has nearly 80k employees to basically run two websites and three mobile apps. There were 18k people working at LinkedIn! Many big tech companies are massive job programs with some product on the side. Administrative business partners, program managers, tech writers, "stewards", "champions", "advocates", 10-layer-deep reporting chains... engineers writing cafe menu apps and pet programming languages... a team working on in-house typefaces... the list goes on.
I can see AI producing shifts in the industry by reducing demand for meaningful work, but I doubt the outcome here is mass unemployment. There's an endless supply of bs jobs as long as the money is flowing.
They build generative AI tools so people can make ads more easily.
They have some of the most sophisticated tracking out there. They have shadow profiles on nearly everyone. Have you visited a website? You have a shadow profile even if you don't have a Facebook account. They know who your friends are based on who you are near. They know what stores you visit when.
Large fractions of their staff are making imperceptible changes to ads tracking and feed ranking that are making billions of dollars of marginal revenue.
What draws you in as a consumer is a tiny tip of the iceberg of what they actually do.
The reality is that you could run LinkedIn with far, far fewer people. You probably need fewer than 100 for core engineering, and likely less than 1,000 overall if you include compliance, sales, and so on - especially since a lot of overseas compliance stuff is outsourced to consulting firms, it's not like you have a team of lawyers in every country in the world.
Before there was so much money in the system, we used to run companies that way. Two decades ago, I worked for a company that had tens of millions of users, maintained its own complex nationwide infra (no AWS back then), and had 400 full-time employees. That made coordination problems a lot easier too. We didn't need ten layers of people and project management because there just wasn't that many of us.
Keeping a website running with high uptime is not the goal. Maximizing revenue and profit is. The extra people aren't waste, they're what drive the incremental imperceptible changes that make these companies profitable.
Is it a better user experience now? Yes. Has it boosted my productivity on this project? Absolutely.
But it still needs a ton of hand holding for anything complicated and I still deal with tons of "OK, this bug is fixed now!" followed by manually confirming a bug still exists.
Haiku 4.5 is already so good it's ok for 80% (95%?) of dev tasks.
It's notably lacking newer models (4.5 Opus, 4.6 Sonnet) and models from Gemini.
LLMs appear to naturally progress in short leaps followed by longer plateaus, as breakthroughs are developed such as chain-of-thought, mixture-of-experts, sub-agents, etc.
I don’t see that ever going away. Humans have learned to trust other humans over a large time scale with rules in place to control behaviour.
Once we know what they can do well and how to get them to do it well, and what they can't, you could say we "trust" them to do the first category well and just stop trying to get it to do the second category.
We've compared now more than a hundred replies to that of GPT Pro, and the quality is roughly the same. Sometimes a little worse, sometimes a little better. Always more detailed. Never unacceptable.
But how to convince our customers that we have the right technology and know how to use it appropriately? We're trying, but it's not easy.
Part of that's accountability. In the event of the LLM producing rubbish, as rare as it may be, who is accountable? There is not a person and her reputation attached to it.
You won’t be able to scale out and make as much money though. But surely you’re not only concerned about profit, right? What’s the point of life if you’re just trying to get rich.
Being able to hold someone liable for a F up has been how we have been able to function as a society and get to where we are today.
In particular GPT 5.4 is much better at not duplicating code unnecessarily. It'll take the time to refactor, to search for pre-existing utility functions, etc.
Most of us have been coding for ages. I actually find it really odd people keep trying to disprove things that are relatively obvious with LLMs
You kind of have to go on "feels" for a lot of this.
But LLMs are hamstrung by their harnesses. They are doing the equivalent of providing technical support via phone call: little to no context, and limited to a bidirectional stream of words (tokens). The best agent harnesses have the equivalent of vision-impairment accessibility interfaces, and even those are still subpar.
Heck, giving LLMs time to think was once a groundbreaking idea. Yesterday I saw Claude Code editing a file using shell redirects! It's barbaric.
I expect future improvements to come from harness improvements, especially around sub agents/context rollbacks (to work around the non-linear cost of context) and LLM-aligned "accessibility tools". That, or more synthetic training data.
entirely so. i think anthropic updated something about the compact algorithm recently, and its gone from working well over long times to basically garbage whenever a compact happens
And search engines are narrow tools that can only output copies of its dataset. An LLM is capable of surprisingly novel output, even if the exact level of creativity is heavily debated.
Assuming we get no better than opus 4.6, they're very capable. Even if they make up nonsense 5% of the time!
It will be wrong on a lot of details. Basically you would get a market scene that feels 18th century gb, but will use 18th century russian/french/Austrian details, or 19th century/20th century British artefacts, or a mix of both. And the further you go from western places, the higher the error rate is. Basically generating fiction scenes. That's pretty much my usecase, so that's fine, but I won't ever use it to illustrate a historical TTrpg.
So I’d say fairly flat commit acceptance numbers make sense even in the context of improving LLMs
Truth is I'm probably wrong. I should keep on testing ... but at the same time I precisely gave up because I didn't think the trend was fast enough to keep on investing on checking it so frequently. Now I just read this kind of post, ask around (mainly arguing with comments asking for genuine examples that should be "surprising" and kept on being disappointed) and that seems to be enough for a proxy.
I should though, as I mentioned in another comment, keep track of failed attempts.
PS: I check solely on self-hosted models (even if not on my machine but least on machines I could setup) because I do NOT trust the scaffolding around proprietary closed sources models. I can't verify that nobody is in the loop.
On the other hand, LLMs tend to go for an average by their nature (if you squint enough). What's more common in their training data, it's more common in the output, so getting them better without fundamental changes, requires one to improve the training data on average too which is hard.
What did improve a lot is the tooling around them. That's gotten way better.
Supposedly model curation is a Big Deal at Big AI, and they're especially concerned about Ouroboros effects and poisoned data. Also people are still contributing to open source and open sourcing new projects, something that should have slowed to trickle by 2023, once it became clear that from now on, you're just providing the fuel for the machines that will ultimately render you unemployable (or less employable), and that these machines will completely disregard your license terms, including those of the most permissive licenses that seek only attribution, and that you're doing all of this for free.
I think the only reasonable thing to read into is Sonnet 3.5 -> 3.7 -> 4.5. But yeah, you just can't draw a line through this thing.
I will die on the hill that LLMs are getting better, particularly Anthropic's releases since December. But I can't point at a graph to prove that, I'm just drawing on my personal experience. I do use Claude Code though, so I think a large part of the improvement comes from the harness.
Regardless I'm more inclined to believe that 4.5 was the point that people started using it after having given up on copy/pasting output in 2024. If you're going from chat to agentic level of interaction it's going to feel like a leap.
I don't know if it's model, or harness improvements, or inbuilt-memory or all of the above, but it often has a step where it'll check itself that is done now before trying to build and getting an inevitable failure.
Those small things add up to a much smoother and richer experience today compared to 6 months ago.
When you combine models with:
tool use
planning loops
agents that break tasks into smaller pieces
persistent context / repos
the practical capability jump is huge.
We still have tons of gaps about how to build and maintain code with AI, but LLM themselves getting better at an unbelievable pace, even with this kind of data analysis I’m surprised anyone can even question it.
How much of that is the model and how much of that is the tooling built around it? Also why is the tooling, specifically Claude Code, so buggy?
It's my experience that opus 4, and then, particularly, 4.5, in Claude code, are head and shoulders above the competition.
I wrote an agentic coder years ago and it yielded trash. (Tried to make it do then what kiro does today).
The models are better. Now, caveat - I don't use anything but opus for coding - Sonnet doesn't do the trick. My experience with Codex and Gemini is that their top models are as good as Sonnet for coding...
Although I feel like for chasing bugs and big systems codex is even better
Without any data from the study past September I think its not unreasonable, if you want to make an argument based on evidence.
For me personally, I agree with you, I'm really seeing it as well.
well, yeah. because that's been the experience for many people.
3 years ago, trying to use ChatGPT 3.5 for coding tasks was more of a gimmick than anything else, and was basically useless for helping me with my job.
today, agentic Opus 4.6 provides more value to me than probably 2 more human engineers on my team would
I tried GPT3.5 for translating code from typescript to rust. It made many mistakes in rust. It couldn't fix borrow checker issues. The context was so small that I could only feed it small amounts of my program at a time. It also introduced new bugs into the algorithm.
Yesterday I had an idea for a simple macos app I wanted. I prompted claude code. It programmed the whole thing start to finish in 10 minutes, no problem. I asked it to optimize the program using a technique I came up with, and it did. I asked it to make a web version and it did. (Though for some reason, the web version needed several rounds of "it doesn't work, here's the console output").
I'm slowly coming to terms with the idea that my job is fundamentally changing. I can get way more done by prompting claude than I can by writing the code myself.
Yes but this blogpost argues that at least over the course of 2024 to the end of 2025, those people were mistaken.
I'm certainly getting faster and cleaner-looking solutions for certain issues on Opus 4.6 than I was 5 months ago, but I'm not sure about the ability to solve (or even weigh in) the actual hard stuff, i.e. the stuff I'm paid for.
And I'm definitely not sure about the supposed big step between 4.5 and 4.6. I'm literally not seeing any.
So we got better in giving it the right context and tools to do the stuff we need to do but not the actual thinking improvements
What's been the game changer are tools like Claude Code. Automatic agentic tool loops purpose built for coding. This is what I have seen as the impetus for mainstream adoption rather than noticeable improvements in ability.
I write a lot of C++ and QML code. Codex 5.3, only released in Feb, is the the first model I've used that would regularly generate code that passes my 25 years expert smell test and has turned generative coding from a timesap/nuisance into a tool I can somewhat rely on not to set me back.
Claude still wasn't quite there at the time, but I haven't tried 4.6 yet.
QML is a declarative-first markup language that is a superset of the JavaScript syntax. It's niche and doesn't have a giant amount of training data in the corpus. Codex 5.3 is the first model that doesn't super botch it or prefers to write reams of procedural JS embeds (yes, after steering). Much reduced is also the tendency to go overboard on spamming everything with clouds of helper functions/methods in both C++ and QML. It knows when to stop, so to speak, and is either trained or able to reason toward a more idiomatic ideal, with far less explicit instruction / AGENTS.md wrangling.
It's a huge difference. It might be the result of very specific optimization, or perhaps simultaneous advancements in the harness play a bigger role, but in my books my kneck of the woods (or place on the long tail) only really came online in 2026 as far as LLMs are concerned.
[1] https://rubymamistvalove.com/block-editor
It's just a point release and it isn't a significant upgrade in terms of features or capabilities, but it works... better for me.
My own experience is that some things get better and some things get worse in perceived quality at the micro-level on each point release. i.e. 4.5->4.6
At the end of the day they still produce code that I need to manually review and fully understand before merging. Usually with a session of back-and-forth prompting or manual edits by me.
That was true 2 years ago, and it’s true now (except 2 years ago I was copy/pasting from the browser chat window and we have some nicer IDE integration now).
This is why you have to reach to things that penalize adding parameters to models when running model comparisons.
That actually helps.
I mean, sure. but it's obvious in that graph that the single openai model is dragging down the right side. Wouldn't it be better to just stick to analyzing models from only one lab so that this was showing change over time rather than differences between models?
There was a long flat line before the step, models improve, but PR pass rate without human intervention is inherently a staircase function
As they said, ragebait used to be believable.
There's only so much data to train on, and we are unlikely to see giant leaps in performance as we did in 2023/2024.
2026-27 will be the years of primarily ecosystem/agentic improvements and reducing costs.
Because it's not true. They have improved tremendously in the last year, but it looks like they've hit a wall in the last 3 months. Still seeing some improvements but mostly in skills and token use optimization.
I have heard rumors that token use optimization has been a recent focus to try to tidy up the financials of these companies before they IPO. take that with a grain of salt though
> OpenAI’s leading researchers have not completed a successful full-scale pre-training run that was broadly deployed for a new frontier model since GPT-4o in May 2024 [1]
That's evidence against "intrinsically better". They've also trained on the entire internet - we only have 1 internet, so.
However, late 2024 was the introduction of o1 and early 2025 was Deepseek R1 and o3. These were definitely significant reasoning models - the introduction of test time compute and significant RL pipelines were here.
Mid 2025 was when they really started getting integrated with tool calling.
Late 2025 is when they really started to become agentic and integrate with the CLI pretty well (at least for me). For example, codex would at least try and run some smoke tests for itself to test its code.
In early 2026, the trend now appears to be harness engineering - as opposed to "context engineering" in 2025, where we had to preciously babysit 1 model's context, we make it both easier to rebuild context (classic CS trick btw: rebooting is easier than restoring stale state [2]) and really lean into raw cli tool calling, subagents, etc.
[1] https://newsletter.semianalysis.com/p/tpuv7-google-takes-a-s...
[2] https://en.wikipedia.org/wiki/Kernel_panic
FWIW, AI programming has still been as frustrating as it was when it was just TTC in 2025. Maybe because I don't have the "full harness" but it still has programming styles embedded such as silent fallback values, overly defensive programming, etc. which are obvoiusly gleaned from the desire to just pass all tests, rather than truly good programming design. I've been able to do more, but I have to review more slop... also the agents are really unpleasant to work with, if you're trying to have any reasonable conversation with them and not just delegate to them. It's as if they think the entire world revolves around them, and all information from the operator is BS, if you try and open a proper 2-way channel.
It seems like 2026 will go full zoom with AI tooling because the goal is to replace devs, but hopefully AI agents become actually nice to work with. Not sycophantic, but not passively aggressively arrogant either.
>To study how agent success on benchmark tasks relates to real-world usefulness, we had 4 active maintainers from 3 SWE-bench Verified repositories review 296 AI-generated pull requests (PRs). We had maintainers (hypothetically) accept or request changes for patches as well as provide the core reason they were requesting changes: core functionality failure, patch breaks other code or code quality issues.
I would also advise taking a look at the rejection reasons for the PRs. For example, Figure 5 shows two rejections for "code quality" because of (and I quote) "looks like a useless AI slop comment." This is something models still do, but that is also very easily fixable. I think in that case the issue is that the level of comment wanted hasn't been properly formalized in the repo and the model hasn't been able to deduce it from the context it had.
As for the article, I think mixing all models together doesn't make sense. For example, maybe a slope describe the increasing Claude Sonnet better than a step function.
I also wonder how much of the jump in early 2025 comes from cultural acceptance by devs, rather than an improvement in the tools themselves.
Perhaps we won't see a phase change like improvement as we did from gpt-2 through to 3 until there is several more orders of magnitude parameters and/or training. Perhaps we will never see it again!
What is getting rapidly better is scaffolding but this seems to be more about understanding and building tools around LLMs than the LLMs themselves improving.
I'm still excited about AI but not constantly hyped to the rafters as some.
I guess at least this person https://www.tomshardware.com/tech-industry/artificial-intell... might disagree. I think already to know what Kubernetes even is requires quite a bit of knowledge. Using a tool that manipulate its configuration files IN PRODUCTION without risking data loss is another ball game entirely.
Because hype makes money.