GPT-5.5

LLMs upend a few centuries of labor theory.

The current market is predicated on the assumption that labor is atomic and has little bargaining power (minus unions). While capital has huge bargaining power and can effectively put whatever price it wants on labor (in markets where labor is plentiful, which is most of them).

What happens to a company used to extracting surplus value from labor when the labor is provided by another company which is not only bigger but unlike traditional labor can withhold its labor indefinitely (because labor is now just another for of capital and capital doesn't need to eat)?

Anyone not using in house model is signing up to find out.

One might argue that it’s not too too different from higher level abstractions when using libraries. You get things done faster, write less code, library handles some internal state/memory management for you.

Would one be uneasy about calling a library to do stuff than manually messing around with pointers and malloc()? For some, yes. For others, it’s a bit freeing as you can do more high-level architecture without getting mired and context switched from low level nuances.

> This quote is more sinister than I think was intended; it likely applies to all frontier coding models. As they get better, we quickly come to rely on them for coding. It's like playing a game on God Mode. Engineers become dependent; it's truly addictive.

What's the worst potential outcome, assuming that all models get better, more efficient and more abundant (which seems to be the current trend)? The goal of engineering has always been to build better things, not to make it harder.

Well, they obviously are going to say that, they have vested interest in OpenAI and thus Nvidia stock price growing.

Also, I honestly can’t believe the 10x mantra is being still repeated.

Assuming that local models are able to stay within some reasonably fixed capability delta of the cutting edge hosted models (say, 12 months behind), and assuming that local computing hardware stays relatively accessible, the only risk is that you'll lose that bit of capability if the hosted models disappear or get too expensive.

Note that neither of these assumptions are obviously true, at least to me. But I can hope!

I feel like most engineers I talk to still haven't realised what this is going to mean for the industry. The power loom for coding is here. Our skills still matter, but differently.

This engineer had their brain amputated once they started using AI. All the AI-addicted can do is tinker with the AI computer game and feel "productive". They could as well play Magic The Gathering.

That's the path we've been going down for a few years now. The current hedge is that frontier labs are actively competing to win users. The backup hedge is that open source LLMs can provide cheap compute. There will always be economical access to LLMs, but the provider with the best models will be able to charge basically whatever they want and still have buyers.

soooooo about Claude going down. we're gonna need you to sign in on Saturday and make up for lost time or unfortunately we're going to have to deduct the time lost from your paycheck. and as an aside your TPS reports have been sub-par as of late..is everything OK?

You are now a manager. If your minions are out sick, project is delayed, not the end of the world.

> It's literally higher leverage for me to go for a walk

Touching grass while you're outside might yield highest leverage.

i wonder if this is how engineers felt when the first electronic calculators came out and engineers stopped doing math by hand.

did we feel uneasy that a new generation of builders didn't have to solve equations by hand because a calculator could do them?

i'm not sure it's the same analogy but in some ways it holds.

You’re still the one that’s controlling the model though and steering it with your expertise. At least that’s what I tell myself at night :)

I haven’t really thought about this before, but you’re right, it feels a bit uneasy for me too.

That's why local models are important.

Of course they aren't alternative to the current frontier model, and as such you cannot easily jump from the later to the former, but they aren't that far behind either, for coding Qwen3.5-122B is comparable to what Sonnet was less than a year ago.

So assuming the trend continues, if you can stop following the latest release and stick with what you're already using for 6 or 9 months, you'll be able to liberate yourself from the dependency to a Cloud provider.

Personally I think the freedom is worth it.

The meta here is to use LLMs to make things simpler and easier, not to make things harder.

Turning tokens into a well-groomed and maintainable codebase is what you want to do, not "one shot prompt every new problem I come across".

"when the tokens run out, I'm basically done working."

Oh stop the drama. Open source models can handle 99% of your questions.

Help. They’re constantly trying to make me try crack cocaine on the front page.

It makes me uneasy because my role now, which is prompting copilot, isn't worth my salary.

Given that it’s so easy, would you still do this same job if paid half as much?

eh this kind of FUD needs to stop because it is kind of normal and expected and in fact good to have relation like this with technology.

Did you guys do anything about GPT‘s motivation? I tried to use GPT-5.4 API (at xhigh) for my OpenClaw after the Anthropic Oauthgate, but I just couldn‘t drag it to do its job. I had the most hilarious dialogues along the lines of „You stopped, X would have been next.“ - „Yeah, I‘m sorry, I failed. I should have done X next.“ - „Well, how about you just do it?“ - „Yep, I really should have done it now.“ - “Do X, right now, this is an instruction.” - “I didn’t. You’re right, I have failed you. There’s no apology for that.”

I literally wasn’t able to convince the model to WORK, on a quick, safe and benign subtask that later GLM, Kimi and Minimax succeeded on without issues. Had to kick OpenAI immediately unfortunately.

Conceivably you could have a public-facing dashboard of the rollout status to reduce confusion or even make it visible directly in the UI that the model is there but not yet available to you. The fanciest would be to include an ETA but that's presumably difficult since it's hard to guess in case the rollout has issues.

Congrats on the release! Is Images 2.0 rolling out inside ChatGPT as well, or is some of the functionality still going to be API/Playground-only for a while?

Looks good, but I’m a little hesitant to try it in Codex as a Plus user since I’m not sure how much it would eat into the usage cap.

Will GPT 5.5 fine tuning be released any time soon?

Great stuff! Congrats on the release!

Are you able to say something about the training you've done to 5.5 to make it less likely to freak out and delete projects in what can only be called shame?

Just a tip: add [translated] subtitles to the top video.

Please next time start with azure foundry lol thanks!

With Anthropic, newer models often lead to quality degradation. Will you keep GPT 5.4 available for some time?

LETS GO CODEX #1

can't wait! Thanks guys. PS: when you drop a new model, it would be smart to reset weekly or at least session limits :)

That's amazing that the default did that much in just 39 "reasoning tokens" (no idea what a reasoning token is but that's still shockingly few tokens)

That pelican you posted yesterday from a local model looks nicer than this one.

Edit: this one has crossed legs lol

Isn't it awful ? After 5.5 versions it still can't draw a basic bike frame. How is the front wheel supposed to turn sideways ?

Thank you for doing all this. It's appreciated.

Is this direct API usage allowed by their terms? I remember Anthropic really not liking such usage.

Hmm. Any idea why it's so much worse than the other ones you have posted lately? Even the open weight local models were much better, like the Qwen one you posted yesterday.

I made pelicans at different thinking efforts:

https://hcker.news/pelican-low.svg

https://hcker.news/pelican-medium.svg

https://hcker.news/pelican-high.svg

https://hcker.news/pelican-xhigh.svg

Someone needs to make a pelican arena, I have no idea if these are considered good or not.

what is your setup for drawing pelican? Do you ask model to check generated image, find issues and iterate over it which would demonstrate models real abilities?

It's... like no pelican I've ever seen before.

I for one delight in bicycles where neither wheel can turn!

It continues to amaze me that these models that definitely know what bicycle geometry actually looks like somewhere in their weights produces such implausibly bad geometry.

Also mildly interesting, and generally consistent with my experience with LLMs, that it produced the same obvious geometry issue both times.

Wait, I thought we were onto racoons on e-scooters to avoid (some of) the issues with Goodhart's Law coming into play.

Exciting. Another Pelican post.

You know they are 1000% training these models to draw pelicans, this hasn't been a valid benchmark for 6 months +

At some point, OpenAI is going to cheat and hardcode a pelican on a bicycle into the model. 3D modelling has Suzanne and the teapot; LLMs will have the pelican.

Doesn't OpenAI get mad if you ask cybersecurity questions and force you to upload a government ID, otherwise they'll silently route you to a less capable model?

> Developers and security professionals doing cybersecurity-related work or similar activity that could be mistaken by automated detection systems may have requests rerouted to GPT-5.2 as a fallback.

https://developers.openai.com/codex/concepts/cyber-safety

https://chatgpt.com/cyber

The real 'hype' was that the oh-snap realization that Open AI would absolutely release a competitive model to Mythos within weeks of Anthropic announcing there's, and that Sam would not gate access to it. So the panic was that the cyber world had only a projected 2 weeks to harden all these new zero days before Sam would inevitably create open season for blackhats to discover and exploit a deluge of zero-days.

> Never thought I'd say this but OpenAI is the 'open' option again.

Compared to Anthropic, they always have been. Anthropic has never released any open models. Never released Claude Code's source, willingly (unlike Codex). Never released their tokenizer.

isnt it like cyber question are being routed to dumper models at openai?

> Anthropic's gated Mythos model

aka the perfect marketing ploy

For API usage, GPT-5.5 is 2x the price of GPT-5.4, ~4x the price of GPT-5.1, and ~10x the price of Kimi-2.6.

Unfortunately I think the lesson they took from Anthropic is that devs get really reliant and even addicted on coding agents, and they'll happily pay any amount for even small benefits.

FWIW I've been experimenting with Three.js and AI for the last ~3 years, and noticed a significant improvement in 5.4 - the biggest single generation leap for Three.js specifically. It was most evident in shaders (GLSL), but also apparent in structuring of Three.js scenes across multiple pages/components.

It still struggles to create shaders from scratch, but is now pretty adequate at editing existing shaders.

In 5.2 and below, GPT really struggled with "one canvas, multiple page" experiences, where a single background canvas is kept rendered over routes. In 5.4, it still takes a bit of hand-holding and frequent refactor/optimisation prompts, but is a lot more capable.

Excited to test 5.5 and see how it is in practice.

A friend is building Jamboree[1] (prev name "Spielwerk") for iOS. An app to build and share games. They're all web based so they're easy to share.

[1] https://apps.apple.com/uz/app/jamboree-game-maker/id67473110...

I’ve had a lot of success using LLMs to help with my Three.js based games and projects. Many of my weird clock visualizations relied heavily on it.

It might not be a game engine, but it’s the de facto standard for doing WebGL 3D. And since it’s been around forever, there’s a massive amount of training data available for it.

Before LLMs were a thing, I relied more on Babylon.js, since it’s a bit higher level and gives you more batteries included for game development.

LLM models can not do spacial reasoning. I haven't tried with GPT, however, Claude can not solve a Rubik Cube no matter how much I try with prompt engineering. I got Opus 4.6 to get ~70% of the puzzle solved but it got stuck. At $20 a run it prohibitively expensive.

The point is if we can prompt an LLM to reason about 3 dimensions, we likely will be able to apply that to math problems which it isn't able to solve currently.

I should release my Rubiks Cube MCP server with the challenge to see if someone can write a prompt to solve a Rubik's Cube.

The meshes look interesting, but the gameplay is very basic. The tank one seems more sophisticated with the flying ships and whatnot.

What's strange is that this Pietro Schirano dude seems to write incredibly cargo cult prompts.

  Game created by Pietro Schirano, CEO of MagicPath

  Prompt: Create a 3D game using three.js. It should be a UFO shooter where I control a tank and shoot down UFOs flying overhead.
  - Think step by step, take a deep breath. Repeat the question back before answering.
  - Imagine you're writing an instruction message for a junior developer who's going to go build this. Can you write something extremely clear and specific for them, including which files they should look at for the change and which ones need to be fixed?
  -Then write all the code. Make the game low-poly but beautiful.
  - Remember, you are an agent: please keep going until the user's query is completely resolved before ending your turn and yielding back to the user. Decompose the user's query into all required sub-requests and confirm that each one is completed. Do not stop after completing only part of the request. Only terminate your turn when you are sure the problem is solved. You must be prepared to answer multiple queries and only finish the call once the user has confirmed they're done.
  - You must plan extensively in accordance with the workflow steps before making subsequent function calls, and reflect extensively on the outcomes of each function call, ensuring the user's query and related sub-requests are completely resolved.

It’s like all these things though - it’s not a real production worthy product. It’s a super-demo. It looks amazing until you realize there’s many months of work to make it something of quality and value.

I think people are starting to catch on to where we really are right now. Future models will be better but we are entering a trough of dissolution and this attitude will be widespread in a few months.

I personally don't think the gameplay itself is that impressive.

There's already KernelBench which tests CUDA kernel optimizations.

On the other hand all companies know that optimizing their own infrastructure / models is the critical path for ,,winning'' against the competition, so you can bet they are serious about it.

Honestly the problem with these is how empirical it is, how someone can reproduce this? I love when Labs go beyond traditional benchies like MMLU and friends but these kind of statements don't help much either - unless it's a proper controlled study!

They mentioned in their release page, that the Claude team noticed memorization of the SWE-bench test, so the test is actually in the training data.

Here: https://www.anthropic.com/news/claude-opus-4-7#:~:text=memor...

Mythos is only real when it's actually available. If you're using Opus 4.7 right now, you know how incredibly nerfed the Opus autonomy is in service of perceived safety. I'm not so confident this will be as great as Anthropic wants us to believe..

I did some study on Verified, not Pro, but Mythos number there rings a lot of questions on my end.

If you look at the SWEBench official submissions: https://github.com/SWE-bench/experiments/tree/main/evaluatio..., filter all models after Sonnet 4, and aggregate ALL models' submission across 500 problems, what I found that the aggregated resolution rate is 93% (sharp).

Mythos gets 93.7%, meaning it solves problems that no other models could ever solve. I took a look at those problems, then I became even more suspicious, for the remaining 7% problems, it is almost impossible to resolve those issues without looking at the testing patch ahead of time, because how drastically the solution itself deviates from the problem statement, it almost feels like it is trying to solve a different problem.

Not that I am saying Mythos is cheating, but it might be too capable to remember all states of said repos, that it is able to reverse engineer the TRUE problem statement by diffing within its own internal memory. I think it could be a unique phenomena of evaluation awareness. Otherwise I genuinely couldn't think of exactly how it could be this precise in deciphering such unspecific problem statements.

A single benchmark is meaningless, you always get quirky results on some benchmarks.

grok is 17%? And that's the lowest, most models are like 80%+?

While hallucination is probably closer to 100% depending on the question. This benchmark makes no sense.

There's something off with this because Haiku should not be that good.

This indicates they want this behavior, they know the person asking the question probably doesn't understand the problem entirely (or why would they be asking), so they'd prefer a confident response, regardless of outcomes, because the point is to sell the technologies competency (and the perception thereof), not the capabilities, to a bunch of people that have no clue what they're talking about.

LLMs will ruin your product, have fun trusting a billionaires thinking machine they swear is capable of replacing your employees if you just pay them 75% of your labor budget.

Do "our [superlative] and [superlative] [product] yet" and you have pretty much every product launch

"our newest and most expensive model yet"

"Best iPhone ever"

can't wait for "our worst and dumbest model yet"

It's genuinely so great at long horizon tasks! GPT-5.5 solved many long-horizon frontier challenges, for the first time for an AI model we've tested, in our internal evals at Canva :) Congrats on the launch!

Could be a great feature, can't wait to test! Tired of other models (looking at you Opus) constantly stuck mid-task lately.

As others have mentioned you're ignoring the long tail of open-weights models which can be self hosted. As long as that quasi-open-source competition keeps up the pace, it will put a cap on how expensive the frontier models can get before people have to switch to self-hosting.

That's a big if, though. I wish Meta were still releasing top of the line, expensively produced open-weights models. Or if Anthropic, Google, or X would release an open mini version.

This is entirely expected. The low prices of using LLMs early on was totally and completely unsustainable. The companies providing such services were (and still are) burning money by the truckload.

The hope is to get a big userbase who eventually become dependent on it for their workflow, then crank up the price until it finally becomes profitable.

The price for all models by all companies will continue to go up, and quickly.

Look a cost per intelligence or cost per task instead of cost per token.

We know they cost much more than this for OpenAI. Assume prices will continue to climb until they are making money.

Not really a big problem. Switch to KIMI, Qwen, GLM. You’ll get 95% quality of GPT or Anthropic for a 10th of a price. I feel like the real dependency is more mental, more of a habit but if you actually dip your toes outside OpenAI, Anthropic, Gemini from time to time, you realise that the actual difference in code is not huge if prompted in a good way. Maybe you’ll have to tell it to do something twice and it won’t be a one shot, but it’s really not an issue at all.

It's far more meaningful to look at the actual cost to successfully something. The token efficiency of GPT-5.5 is real; as well as it just being far better for work.

SOTA models get distilled to open source weights in ~6 months. So paying premium for bleeding edge performance sounds like a fair compensation for enormous capex.

Such an increase tracks the company's valuation trend, which they constantly, somehow have to justify (let alone break even on costs).

Start building your own liteweight "harness" that does things you need. Ignore all functionality of clients like CC or Codex and just implement whatever you start missing in your harness.

You can replace pretty much everything - skills system, subagents, etc with just tmux and a simple cli tool that the official clients can call.

Oh and definitely disable any form of "memory" system.

Essentially, treat all tooling that wraps the models as dumb gateways to inference. Then provider switch is basically a one line config change.

It's surprisingly simple to switch. I mean both products offer basically identical coding CLI experiences. Personally I've been paying for Claude max $100, and ChatGPT $20, and then just using ChatGPT to fill in the gaps. Specifically I like it for code review and when Claude is down.

I have a directory of skills that I symlink to Codex/Claude/pi. I make scripts that correspond with them to do any heavy lifting, I avoid platform specific features like Claude's hooks. I also symlink/share a user AGENTS.md/CLAUDE.md

MCPs aren't as smooth, but I just set them up in each environment.

Anecdotally, I get the same wall time with my Max x5 (100$) and my ChatGPT Teams (30$) subscriptions.

What is the switching cost besides launching a different program? Don’t you just need to type what you want into the box?

This might be the opposite of staying nimble as my workflows are quite tied to Claude Code specifically, however I've been experimenting with using OpenAI models in CC and it works surprisingly well.

Small tip, at least for now you can switch back to Opus 4.6, both in the ui and in Claude Code.

It’s good to just keep trying different ones from time to time.

Except for history, I don’t find much that stops you from switching back and forth on the CLI. They both use tools, each has a different voice, but they both work. Have it summarize your existing history into a markdown file, and read it in with any engine.

The APIs are pretty interchangeable too. Just ask to convert from one to the other if you need to.

As a rule I've been symlinking or referencing generic "agents" versions of claude workflow files instead of placing those files directly in claude's purview

AGENTS.md / skills / etc

I switched a couple of weeks ago just to see how it went. Codex is no better or worse. They’re both noticeably better at different things. I burn through my tokens much much faster on Codex though. For what it’s worth I’m sticking with Codex for now. It seems to be significantly better at UI work although has some really frustrating bad habits (like loading your UI with annoying copywriting no sane person would ever do).

Coding models are effectively free. They are capable of making money and supporting themselves given access to the right set of things. That is what I do

I've come to prompt pelicans and chew gum, and I'm all outta gum!

That's a true CTO right there.

I know a 10x engineer when i see one.

Ctrl+F: pelican

F5

simonw pls

Clearly they felt a big backlash when version 5 was released. Now they are afraid of another response like this. And effectively, for the user it will likely only be a small update.

Also the naming department. You can tell that this is the AI company Microsoft chose to back because their naming scheme is as bad as .NET's.

As someone who always leaves intelligence at default, and am ok with existing models, should I be shifting gears more manually as providers sell us newer models? Is medium or lower better than free/cheaper models?

Well anectodally, 5.4 was already better than opus 4.7 so it should not have been hard.

I like that Anthropic rushed 4.7 out to get a couple days of coverage before 5.5 hit

why would chip affect token quantity. this is all models.

You need to compare total cost. Token count is irrelevant.

Chips doesn’t impact output quality in this magnitude

If it's a new pretrain, the token embeddings could be wider - you can pack more info into a token making it's way through the system.

Like Chinese versus English - you need fewer Chinese characters to say something than if you write that in English.

So this model internally could be thinking in much more expressive embeddings.

Yep, it's more expensive per token.

However, I do want to emphasize that this is per token, not per task.

If we look at Opus 4.7, it uses smaller tokens (1-1.35x more than Opus 4.6) and it was also trained to think longer. https://www.anthropic.com/news/claude-opus-4-7

On the Artificial Analysis Intelligence Index eval for example, in order to hit a score of 57%, Opus 4.7 takes ~5x as many output tokens as GPT-5.5, which dwarfs the difference in per-token pricing.

The token differential varies a lot by task, so it's hard to give a reliable rule of thumb (I'm guessing it's usually going to be well below ~5x), but hope this shows that price per task is not a linear function of price per token, as different models use different token vocabularies and different amounts of tokens.

We have raised per-token prices for our last couple models, but we've also made them a lot more efficient for the same capability level.

(I work at OpenAI.)

That pricing is extremely spicy, wow.

yes but as far as i know gpt tokenizer is about the same as opus 4.6's, where 4.7 is seeing something in the ballpark of a 30% increase. this should still be cheaper even disregarding the concerns around 4.7 thinking burning tokens

The premise of the trillion dollars in AI investments is not that it’ll be as good as it currently is but cheaper. It’s AGI or bust at this point.

Isn't Mythos limited to a selected group of companies/organizations Anthropic chose themselves? If the OpenAI announcement for GPT-5.5 is accurate the "trusted cyber access" just requires an open, seemingly straightforward identity verification step.

https://openai.com/index/scaling-trusted-access-for-cyber-de...

  > We are expanding access to accelerate cyber defense at every level. We are making our cyber-permissive models available through Trusted Access for Cyber , starting with Codex, which includes expanded access to the advanced cybersecurity capabilities of GPT‑5.5 with fewer restrictions for verified users meeting certain trust signals (opens in a new window) at launch.

  > Broad access is made possible through our investments in model safety, authenticated usage, and monitoring for impermissible use. We have been working with external experts for months to develop, test and iterate on the robustness of these safeguards. With GPT‑5.5, we are ensuring developers can secure their code with ease, while putting stronger controls around the cyber workflows most likely to cause harm by malicious actors.

  > Organizations who are responsible for defending critical infrastructure  can apply to access cyber-permissive models like GPT‑5.4‑Cyber, while meeting strict security requirements to use these models for securing their internal systems.

"GPT‑5.4‑Cyber" is something else and apparently needs some kind of special access, but that CyberGym benchmark result seems to apply to the more or less open GPT-5.5 model that was just released.

Isn't CyberGym an open benchmark so trivial to benchmaxx anyway?

Not good for employees that are being measured by their token usage.

I was all in on Claude code as my daily driver for web development. And love it. But I enjoy using pi as my harness more and have never ran out of tokens with Codex yet. Claude code almost always runs out for me with the same amount of usage.

After migrating for the token and harness issues, I was pleasantly surprised that Codex seems to perform as good or better too!

Things change so often in this field, but I prefer Codex now even though Anthropocene has so much more hype for coding it seems.

I switched from CC to Codex a few days ago. I get limited much less and the code quality is similar, so not looking back

I have experienced 0 friction swapping between the 2 models, in fact pitting them against eachother has resulted in the highest success rate for me so far.

Every time I've followed the hype and tried OpenAI models I've found them lacking for the most part. It might just be that I prefer the peer-programming vs spec-ing out the task and handing it off, but I've never been as productive as I am with Claude. Also, I'm still caught up on the DoD ethics stuff.

I’ve noticed when writing little bedtime stories that require specific research (my kids like Pokemon stories and they’ve been having an episodic “pokemon adventure” with them as the protagonists) ChatGPT has done a fantastic job of first researching the moves the pokemon have, then writing the actual story. The only mistake it consistently makes is when I summarize and move from a full context session, it thinks that Gyarados has to swim and is incapable of flying.

It definitely seems like it does all the searching first, with a separate model, loads that in, then does the actual writing.

It's literally a distinct model with a different optimisation goal compared to normal chat. There's a ton of public information around how they work and how they're trained

It's a property of the model in the sense that it has great Google Fu.

The harness provides the search tool, but the model provides the keywords to search for, etc.

For less than 10% bump across the benchmarks? Probably not, but if your employer is paying (which is probably what OAI is counting on) it's all good.

It's kind of starting to make sense that they doubled the usage on Pro plans - if the usage drains twice as fast on 5.5 after that promo is over a lot of people on the $100 plan might have to upgrade.

Gpt-5.3-codex is miles better than 5.4 in that regard. It’s better at orchestration, and does the things that it said it did. Haven’t tested 5.5 yet but using 5.4 for exploration + brainstorming and handing over the findings to 5.3-codex works pretty well

The only difference being that we still do care, very much. The models can still get a lot better before we stop caring.

It's behind Opus 4.7 in SWE-Bench Pro, if you care about that kind of thing. It seems on-trend, even though benchmarks are less and less meaningful for the stuff we expect from models now.

Will be interesting to try.

Use codex app

After 5.1, we haven’t seen a -codex-max model, presumably because the benefits of the special training gpt-5.1-codex-max got to improve long context work filtered into gpt-5.2-codex, making the variant no longer necessary (my personal experience accords with this). I’ve been using gpt-5.4 in Codex since it came out, it’s been great. I’ve never back-to-back tested a version against its -codex variant to figure out what the qualitative difference is (this would take a long time to get a really solid answer), but I wouldn’t be surprised if at some point the general-purpose model no longer needs whatever extra training the -codex model gets and they just stop releasing them.

I thought it was weird that for almost the entire 5.3 generation we only had a -codex model, I presume in that case they were seeing the massive AI coding wave this winter and were laser focused on just that for a couple months. Maybe someday someone will actually explain all of this.

With Opus it’s hard to tell what was due to the tokenizer changes. Maybe using more tokens for the same prompt means the model effectively thinks more?

They say latency is the same as 5.4 and 5.5 is served on GB200 NVL72, so I assume 5.4 was served on hopper.

To be fair, there's not much to report. Isn't it pretty much at 0?

GPT-5.5 was just released and OpenAI didnt mention ARC AGI 3 at all, their score probably sucks.

I don't think this is technically possible without something like homomorphic encryption, which poses too large of a runtime cost for usage in LLMs

Is it? There are 5 other models near ~80% and it was achieved in March... which in AI-world seems like a century ago.

https://www.tbench.ai/leaderboard/terminal-bench/2.0

Competition.

Anthropic is really tiny, and Google is just being Google, their models are just to show that they're hip with what the kids are doing.

I wonder if it's the same model and they just keep adding more post-training.

They aren't new models.

GPT 5.4 is already better than Opus 4.7 to me. But, then again, Opus 4.7 is a massive disappointment. I hope they don't discontinue 4.6.

There's an asterisk right below that table stating that:

> *Anthropic reported signs of memorization on a subset of problems

And from the Anthropic's Opus 4.7 release page, it also states:

> SWE-bench Verified, Pro, and Multilingual: Our memorization screens flag a subset of problems in these SWE-bench evals. Excluding any problems that show signs of memorization, Opus 4.7’s margin of improvement over Opus 4.6 holds.

Was 4.7 distilled off Mythos (which got 77.8%)? Interesting how mythos got 82% on terminal-bench 2.0 compared to 82.7% for GPT-5.5.

Also notice how they state just for SWE-Bench Pro: "*Anthropic reported signs of memorization on a subset of problems"

Almost nobody can actually use Mythos, though?

Hopefully not, because checking your codebase for vulnerabilities is really valuable.

I hope it’s just limits on pentesting and stuff, and not for code analysis and review.

Did you upgrade the tool binaries? I also couldn't see it until after the upgrade.

Nothing. The next major increment is going to happen when marketing department is confident they can sell it as a major improvement without everyone laughing at them. Which at this point seems like never.

I think Anthropic fearmongering and "leaks" of Mythos was them testing the ground for 5.x, which seems to have backfired.

> "information technology" generally didn't increase productivity

Do you think it'd be viable to run most businesses on pen and paper? I'll give you email and being able to consume informational websites - rest is pen and paper.

Its quite possible the use of LLMs means that we are using less effort to produce the same output. This seems good.

But the less effort exertion also conditions you to be weaker, and less able to connect deeply with the brain to grind as hard as once did. This is bad.

Which effect dominates? Difficult to say.

Of course this is absolutely possible. Ultimately there was a time where physical exertion was a thing and nobody was over-weight. That isn't the case anymore is it.

Downvoted by the AI Nazis. They are running a tight ship before the IPOs.

Not this time, no.

Usually, those get released a few weeks later.

I doubt this is representative of real world usage. There is a difference between a few turns on a web chatbot, vs many-turn cli usage on a real project.

This is not any better of a benchmark

The benchmarks say so, but try it out with actual tasks and be the judge.

It's a silly benchmark anyways.