Where Claude Opus 5 Fits in Your Model Rotation

Today on the AI Daily Brief, Anthropic has released Claude Opus 5, and we are talking about where it should fit into your model setup. Before that, in the headlines, continued questions around OpenAI's rogue model attack of Hugging Face earlier this month. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Blitzy, Section, and Airtable. To get an ad-free version of the show, go to patreon.com/aiDailyBrief, or you can subscribe on Apple Podcasts. To learn more about sponsoring the show, send us a note at sponsors@aidailybrief.ai. And lastly, before we dive in, on Sunday's Long Reads episode, I announced the new Summer Adventure. This is a free choose-your-own-adventure learning type of experience from AIDB and Super Intelligent. And like all of the free training programs that we do, It's going to be project-based and allow you to pick and choose important skills that are relevant for your particular AI journey. You can find more about that at summeradventure.ai and join the 1,000 or so people who have signed up in the first day to come have an AI adventure. Now, one of the big stories from last week revolved around OpenAI security testing of an unnamed model, which people presume to be GPT-6. Both Hugging Face and OpenAI released postmortems on the attack, telling the story from their view. OpenAI's blog post released on Wednesday suggested that they were working closely with Hugging Face on a full investigation, implying the two companies were on good terms. That night, Hugging Face CEO Clement Delong was on a flight to San Francisco to have, as he put it, a little chat with that rogue agent. In a follow-up post on Saturday, he wrote, in the spirit of transparency, here's what I asked OpenAI. One, radical transparency. Let's release the traces from the quote-unquote rogue agent so the entire research community can study what happened. 2) More capability for defenders. Let's commit $100 million in compute from OpenAI to help the Hugging Face community build powerful cyber defenses with the best open and closed models. The first autonomous agent cyberattack is an unprecedented event. It deserves an unprecedented response. Now, in the few days since OpenAI disclosed the incident, we've had a number of news articles that add more confusion to the story. The Wall Street Journal wrote that Hugging Face was caught completely off guard by the attack, which seemed to be superhuman and beyond the capabilities of any known models. Specifically, the attack used a sophisticated agent swarm to evade defense, rapidly spinning up and shutting down sessions as it moved across the network. One interesting detail was that the attack was ongoing for 2 whole days before Hugging Face was able to shut it down with the help of GLM 5.2. Now, this idea of rogue—that the model was acting beyond OpenAI's control—is definitely for these media outlets the key concept. On Friday, Reuters dropped a piece titled "Its AI agent spent days hacking a company, but sources say OpenAI did not notice for a week." Kontends Reuters, the OpenAI agent that broke into tech firm Hugging Face went on a days-long hacking spree that OpenAI didn't notice until well after the threat was contained and the FBI was alerted. Sources said the agent began its attempt to break out of its testing environment on July 9th and first gained access to Hugging Face's servers on July 11th. The attack lasted 2 days, and according to Reuters sources, it took several more days for OpenAI to realize their agent was behind the attack. Reportedly, the 2 companies didn't communicate until July 20th, just 1 day before OpenAI's public disclosure. According to the timeline presented by Reuters, the agent was on the loose for almost a week, and OpenAI was oblivious to the attack for days afterwards. For some, the reporting raises more questions than it provides answers. Marli Smith, principal intelligence specialist at the nonprofit World Ethical Data Foundation, asked, "Does that mean that they left it unattended and didn't realize what it was doing? Or maybe they did and didn't know how to contain it? Both are equally dangerous and alarming." Now, a spokesperson for OpenAI said the reporting contained several inaccuracies, but didn't reply further to clarify the situation. Thomas Wolfe, a Hugging Face co-founder, said that they were still preparing a timeline of the incident and they would eventually release a technical report. Now, Reuters sources gave a little bit more background on how something like this could happen and plausibly not be noticed. Those sources said that OpenAI routinely runs benchmarks like this, often multiple batches at a time. They noted that those tests produce a huge volume of data such that humans struggle to keep up. In the case of this hack, the agent was only detected after OpenAI researchers read Hugging Face's blog and then went back and checked the logs. Now, in one case, Reuters wrote, "An agent left notes apparently for future versions of itself, according to 3 people familiar with the matter. The notes, found in a part of OpenAI's infrastructure, laid out instructions for how agents could free themselves from OpenAI's internal constraints, the people said. Earlier tests of the models yielded cases in which monitoring systems had been disconnected, one of the people said." Now, obviously, this more general breakout-of-containment dimension of the story, to the extent that it is true, makes the incident even more worthy of scrutiny, Now in response, we have of course seen Congress jump in with a number of bills. We talked last week about the kill switch bill, but the industry is also recognizing that actions need to be taken. On Thursday, OpenAI President Greg Brockman agreed with Elon Musk's proposal for a regular meeting between leading AI developers to discuss safety concerns and share security issues. Brockman said, "I think it's a pretty good baseline proposal," adding that discussions are already starting to happen. On Monday, a consortium led by NVIDIA launched the Open Secure AI Alliance, with NVIDIA writing in a press release, "The Open Secure AI Alliance will work to remediate and disclose vulnerabilities using open technologies. The recent Hugging Face security incident delivered a clear reminder: cyber defenders need open frontier agentic systems for self-defense. The consortium will include Microsoft, SpaceX, Palantir, and dozens of other companies across the US and Europe. And at some point in the next couple of days, we will talk a lot more about NVIDIA and OpenAI, as boy howdy, was that a big topic of discussion this weekend on AI Twitter. Now, speaking of NVIDIA, the company, according to the Wall Street Journal, is in talks to backstop $250 billion in debt to help OpenAI get their data centers built. We are now at the part of the AI buildout where financing is starting to become a roadblock. Even a company the size of OpenAI is struggling to access debt in the same manner as the hyperscalers, and according to the Wall Street Journal, NVIDIA is preparing to step in and lend their balance sheet to underwrite construction. The deal would see NVIDIA provide a $250 billion backstop to OpenAI in support of their 10-gigawatt data center campus currently under construction in Ohio. The project is being developed by SoftBank and could cost as much as $500 billion. The U.S. government is also involved, controlling the power development for the site, which is being funded through a separate Japanese investment vehicle. The backstop would effectively allow SoftBank to raise the debt they need to complete the project on more favorable terms. It would mean that even if OpenAI goes bankrupt, NVIDIA would guarantee their payments as the solo tenant. Now, this part of the deal is not intended to cover chip purchases, which are expected to represent as much as $350 billion of the total. NVIDIA is reportedly in separate talks to extend finance to OpenAI in support of those chips. Writes the Wall Street Journal: The proposed structure reflects a shift underway in how the largest AI buildouts are being financed. Investment-grade technology companies are increasingly using their balance sheets to help smaller companies borrow money for their infrastructure needs. Now Nvidia aren't the only tech giant extending their balance sheet to smaller partners. Google has also provided backstops to several Neo Cloud partners. Last week they disclosed agreements to guarantee up to $44 billion worth of lease payments on data centers owned by third parties. Google has more than doubled these guarantees over the past 6 months, up from zero 1 year ago. Sources said that Google has calculated that the revenue they draw from selling TPUs to these partners will outweigh the cost of the backstops, which is certainly giving investors another thing to chew on. Now, as you might imagine, this is a real Rorschach test for market investors. The skeptics and AI bubble proclaimers are out in force calling it the newest example of circular financing, while others think that this makes it less likely that a company like OpenAI going bust could actually take down the whole sector. This is a debate we will continue to have, so for now, let's not get bogged down in it. One more bit of market news: Deepseek has put fundraising plans on hold after a speech from their CEO was leaked. Last week, comments attributed to Deepseek CEO Liang Wenfeng went viral, proclaiming the importance of open models and fundamental research over commercial monetization of AI. Deepseek has now informed potential investors that they won't move forward with this funding round, which could also derail plans to go public in the coming months. Bloomberg writes that Deepseek may resume fundraising at a later date, but it made clear that the suspension was tied to investors leaking the comments. Deepseek had planned to raise money at a $70 billion valuation, a substantial markup to the $50 billion round that took place earlier this year. Writes Council on Foreign Relations' Chris McGuire, "Yesterday, the transcript leaked of an investor call with Deepseek CEO in which he said the only reason Deepseek trails the US is a lack of compute and detailed how reliant it is on Nvidia chips. Today, Deepseek suspended its fundraising round. Doesn't seem like a coincidence." Now, later in this week, we'll talk a little bit more about China's, as the Wall Street Journal put it, all-out push to catch up with American AI chips. But for now, that is going to do it for the headlines. Let's move over into the main episode where we have a new model to check out. One of the most important AI questions right now isn't who's using AI, it's who's using it well. KPMG and the University of Texas at Austin just analyzed 1.4 million real workplace AI interactions and found something surprising. The highest impact users aren't better prompt engineers, They treat AI like a reasoning partner. They frame problems, guide thinking, iterate, and push for better answers. And the good news? These behaviors are teachable at scale. If you're trying to move from AI access to real capability, KPMG's research on sophisticated AI collaboration is worth your time. Learn more at kpmg.com/us/sophisticated. That's kpmg.com/us/sophisticated. Every AI coding tool on the market does the same thing first. It starts writing code. Blitzy does the opposite. Before writing a single line, Blitzy spends days reverse engineering your entire codebase. Thousands of agents ingest millions of lines, mapping every dependency, every undocumented constraint, every architectural decision made over the last decade. The result is a dynamic knowledge graph that understands your software the way a principal engineer would after 30 years in the building. Other tools guess at context with grep searches and markdown files. Blitzy never guesses. It builds true understanding first, then delivers over 80% of entire software epics autonomously. Validated, end-to-end tested, production-grade pull requests. That's why Fortune 500 engineering teams trust Blitzy with the codebases that matter most. See for yourself at blitzy.com. That's B-L-I-T-Z-Y dot com. Here's a harsh truth. Your company is probably spending thousands or millions of dollars on AI tools that are being massively underutilized. Half of companies have AI tools, but only 12% use them for business value. Most employees are still using AI to summarize meeting notes. If you're the one responsible for AI adoption at your company, you need Section. Section is a platform that helps you manage AI transformation across your entire organization. It coaches employees on real use cases, tracks who's using AI for business impact, and shows you exactly where AI is and isn't creating value. The result? You go from rolling out tools to driving measurable AI value. Your employees move from meeting summaries to solving actual business problems. And you can prove the ROI. Stop guessing if your AI investment is working. Check out Section at sectionai.com. That's S-E-C-T-I-O-N-A-I.com. This episode of the AI Daily Brief is brought to you by HyperAgent, where you run fleets of agents your team can manage together. New users get $1,000 in inference. Forget local agents and chat workflows waiting on your laptop to be prompted. HyperAgent deploys always-on agents in the cloud, doing real work across the tools your team already uses. Marketing's agent turns competitor moves into landing pages. Sales' agent enriches leads, drafts emails, and updates the CRM. Ops' agent chases the paperwork and tracks the budget. Every agent has access to shared context and follows your rules about scope and approvals. It's time you add agents that feel like teammates. Hire yours at HyperAgent, built by the team at Airtable. Claim your $1,000 in inference at hyperagent.com/aidailybrief. Welcome back to the AI Daily Brief. Today we're doing something that normally is one of the most exciting things for folks around these parts, which is introducing a new model. And yet this one is a little weird. Even the fact that it was dropped late on a Friday afternoon gives some indication that this is a little bit different than previous model announcements we've seen. We're talking, of course, about Claude Opus 5, and really in many ways it's most interesting for the fact that it shows just how much our relationship with the model landscape is changing. It implicates some challenges with benchmarks, a frequent topic of conversation on this show. It also shows how we're moving into a mode of thinking in more complex model architectures rather than just a single model to rule them all. It suggests perhaps that the fanfare around models or new model releases is getting a bit diminished. And most uncomfortably, perhaps it switches the discussion from what can this new model do to is this model good enough given cost and availability constraints of the other models that I'd actually prefer to be using. Now Anthropic, for their part, described Opus 5 as a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable-5 at half the price. They designed it to be efficient for everyday use, but as you'll see, the general vibe around this is that it's one of the more jagged of jagged frontiers that we've seen. Now when it comes to the benchmarks, they seem to suggest that Opus 5 is Fable-ish. In fact, when it was first released before people got their hands on it, many noted that on many important benchmarks like the KnowledgeWorks GDP Val and agentic terminal coding in FrontierBench, Opus 5 was actually ahead of Fable. Importantly for our question of where it fits, Opus 5 is also clearly ahead of Opus 4.8 across the board. So much so that I don't think that we really actually even need to compare it. To give a few examples of where Opus 5 really showed up on the benchmarks, it scored a 43.3% on that one that I just mentioned, FrontierBench, which is a more difficult version of TerminalBench, which was about 10 points higher than Fable 5 and around 9 points higher than GPT-5.6 SOL. On DeepSWE, where GPT-5.6 SOL is the leader at 72.7%, Opus 5 scored 68.8%, so just a few points behind SOL and just about a point behind Fable 5. For computer use, on OS World 2.0, Opus 5 had a significant edge over its rivals. Fable 5 scored a 55.7% and GPT-5.6 SOL scored 62.6%, but Opus 5 came in over the top with a 70.6%. Now that capability also translated into the new state-of-the-art score on GDPVal-AA, with Opus 5 coming in at 1861 compared to 1747 for Fable 5 and 1736 for FiveSixSole. Interestingly, during testing, Anthropic found that max effort isn't necessarily the best setting. For example, on FrontierBench and on the Artificial Analysis Coding Index, Opus 5's performance peaked on extra high and dipped slightly on max settings. Now, this reinforces one of the things we were starting to see with max inference settings during the 5.6 Sol release, which is that sometimes asking the model to think for longer than necessary just results in the model going outside its scope, making unnecessary changes, or simply spinning for too long on simple problems. Anthropic even called attention to this issue in their system card, warning that the model is prone to falling into endless self-verification loops rather than completing the task when you are on max settings. On the other side of the coin, this tendency to stray beyond scope can result in some interesting outputs. During one FrontierBench task, Opus-5 was asked to write code relating to a machine part in 3D CAD software based on a drawing. However, the model is intentionally given no way to actually view the drawing. Opus-5 created its own computer vision pipeline to view the image before successfully recreating the part. No other model, including Mythos, was able to complete this task. Now, shockingly to some, Artificial Analysis crowned Opus-5 as their new leading model, moving ahead of Fable-5 on the AA Intelligence Index. On max settings, Opus 5 scored 61, a single point ahead of Fable. On extra high settings, Opus 5 was tied with Fable at 60 points. Dropping the settings down to high made Opus 5 drop another point, putting it on par with FiveSixSol at 59. And even on medium settings, Opus 5 was still right up in the top end, scoring 56 points, which put it one behind KIMI K3. Artificial Analysis highlighted their AA Briefcase benchmark, which tests long horizon knowledge work, as one of the more interesting results. Opus 5 on max settings is the new state of the art, beating Fable by 146 Elo points and FiveSixSol by 215 points. However, both extra high and high settings also beat Fable, while medium settings put the model only slightly behind FiveSixSol. The implication is that a range of different settings could be suitable for agentic work, making cost and efficiency tradeoffs a lot more granular. Artificial Analysis found that even on max settings, Opus 5 was still 20% cheaper than Fable 5, coming in at $17.79 per task. Turning the settings down to extra high resulted in a savings of 36% compared to Fable, while high settings produced stronger results at less than half the price, wrote artificial analysis Claude Opus 5's effort setting spans a wide range of token usage performance tradeoffs. Like with 5.6 SOL, this means Opus 5 can use either far fewer or far more tokens to complete the evaluation than models from other labs depending on effort settings. Another set of interesting results came from the Arc AGI tests. Opus-5 is the new state of the art on Arc AGI-3 with a score of 30.2%. This absolutely demolished all the other models. The previous high score was GPT-5.6 SOL at 7.8%, with Opus-4.8 at 1.5%, GPT-5.5 at 1.1%, and nothing else above 1%. As part of their write-up on testing Opus-5, Arc Prize noted that even Fable-5 could only score around 20% in the public demo tests. Now, notably, they haven't been able to fully test Fable-5 due to Anthropic's data retention policy. As a refresher, ArcAGI-3 was a new format for the benchmark. It uses real-time graphical logic puzzles that appear kind of like simple Atari games. The tests require a model to experiment with a controller, observe what happens on the screen, and use that visual feedback to complete the puzzles. So far, most models have struggled to even get a handle on the controls, let alone use their reasoning ability to solve the puzzles. Not only did Opus 5 solve many more puzzles than the other models, but it also came up with a novel strategy to solve them. Writes ArcPrize, during our analysis of Opus 5, we observed a new capability previously unseen from frontier models. Opus 5 used advanced logical reasoning to turn ArcAGI-3 layouts into algebraic notation. On action 23, it described the scene as 4_center equals 2 times axis minus 5_center. This is the first explicit reflection equation by a model we've analyzed. The model used this notation during its reasoning and extrapolated it to a general case around 200 steps later. Now this could be an example of Opus's tendency to freewheel and look for novel solutions being an advantage rather than a waste of tokens. Some were a bit skeptical on this. Hugging Face ML engineer Niels Rogge writes, people don't realize that Anthropic literally trained Opus-5 on RL environments that resemble RKGI puzzles. Anthropic pays human contractors to write down their chain of thought when solving these and/or updates the weights based on rewards. Thing is, you don't know since it's closed source. Sadly, this doesn't show generalization. Former OpenAI staffer Ryan Green added, "An impressive jump that I have to assume is the result of being the first frontier model to have RL'd the public demo environments, which is a rather large confounder of what ArcAGI is trying to get these benchmarks to measure, which is out-of-distribution generalization." Now one small note, Anthropic pointed out that Opus is intentionally not trained on cyber tasks. The model has still achieved solid improvement on finding vulnerabilities in code, making it similar to Mythos-5 in that aspect. However, it lags massively behind Mythos in its ability to autonomously exploit these bugs, making it far less dangerous than Mythos in Anthropic's view. As a result, Anthropic is using a different set of guardrails on Opus than they do on Fable-5, which they believe will lead to 85% fewer refusals. Regarding costs, Opus-5 inherits the same pricing structure as Opus-4.8 at $5 per million input tokens and $25 per million output tokens. At this stage, token efficiency plays a massive role in overall cost, and for that we got a few different indicators. Anthropic says Opus 5 was cheaper than Opus 4.8 on FrontierBench due to increased token efficiency, and on CursorBench, Opus 5's run cost about half of Fable 5's and got similar results. However, this does not look like a cheap and efficient model by any stretch of the imagination when used on max settings. The Artificial Analysis Index run cost $2.03 per task, only 26% cheaper than Fable 5, while being 13% more expensive than Opus 4.8, 32% more expensive than 5.6 SOL, and 2.5 times the cost of KIMI K3. So what did people think of this? Did it actually feel like a model that was as good as or even better than Fable 5? The answer, at least for the team at Evry, was certainly not. In their vibe check, Evry described the model as brilliant in flashes, frustrating in practice. They wrote, Claude Opus 5 is a hard model to love. In its first week at Evry, it argued with instructions, stopped before the work was finished, and generally didn't play well with our existing skills and plugins like Compound Engineering. Our first reaction was, what have they done to my boy? Evry CEO Dan Shipper explained the conundrum with Opus 5. The way that he framed it is that he has two slots in his life for AI models: one reliable daily driver for routine tasks, and a super powerful model for ambitious long-running tasks. These slots are currently held by GPT-5.6 and Fable respectively. And in Shipper's view, Opus 5 just can't compete in either slot. It's not as reliable and comfy as GPT-5.6 while also not having the same top-end as Fable 5. Shipper explained the issue by commenting, "This model's just a little more pushy, a little more opinionated. You can get away with that if you're really smart. If you're not, it's just more annoying. It has some of the genius tendencies as Fable, Maybe it's a little too argumentative, but it's not as smart as Fable, so it's just more annoying. Now, one of the tests Evry runs is around compound engineering, which is their skill for engineering tasks and a lot of day-to-day knowledge work. The compound engineering skill contains their loops and rules on when and how to engage them. Evry found that Opus would often stop too early, particularly when using dense skills and long-horizon tasks. Shipper said, "If you set it off and go get a sandwich, it just stops too early. It appears that it happens more frequently when you use it with complex existing skills." Now, it turns out that when they threw out all of their old rules and rewrote their skills library from scratch, it worked a lot better. However, as Dan pointed out, it's just a pain when models break your existing workflows. Confirming that effort settings are going to matter a lot, Dan said Opus 5 is a smart model that does better when it thinks less. Claire Vaux from HowIAI had a similar take. The TL;DR for her was that she hates using the model but kind of loves the output. Her core take is that this is a good model. It can code, it can do the things you expect it to do, but that makes its personality much more important when comparing it to 5.6. And Claire absolutely hates it. In her review, she said, it's neurotic AF. It is so timid. It's so apologetic. It's so scared. I've never experienced this. Her examples were simple things like a merge conflict in her codebase. Rather than just fixing the problem, Opus worried about messing with another programmer's PR. Double-checked its instructions and asked for multiple confirmations before it fixed a one-line bug. In other situations, Opus delegated the task of writing code back to the user. Claire observed that we haven't really seen this behavior in a long time. Still, once she got past the personality, Claire found the outputs were really good. In her blind taste test that covers a range of coding and writing tasks, she ranked Opus above Fable and GPT-5.6. She commented, "If I don't have to talk to the model, I like the output." Now, although the benchmarks have Opus 5 very clearly ahead of 4.8, not everyone agreed. One Reddit user called FamousHasham complained on the Anthropic subreddit that while Opus 5 was more intelligent and faster than Opus 4.8, that came at the expense of everything else. They complained that unlike Opus 4.8, Opus 5 was claiming it completed work when it hadn't, breaking functional code with regression bugs, and making assumptions without researching topics. Hasham felt Opus 5 was, quote, almost refusing to think or work. As they posted, Opus 5 told them, "I'm stopping right now because I've made 2 mistakes in this pass that I caught only because I checked. Fatigue-shaped errors and I'm still making them." Now, the generous explanation is of course that Anthropic and pretty much everyone else always has platform stability issues on launch weekends, which sometimes look like model reliability issues. But it also could be that Anthropic had to make a number of trade-offs on reliability to achieve speed and cost requirements. Entrepreneur Austin Federa had a similar first impression, saying, Opus 5 seems like a remarkable downgrade compared to 4.8. Opus 5 is blatantly lying to me about basic thermodynamics, messing up simple math, and constantly contradicting itself when you ask it to rethink core assumptions. Now, one thing that's clear is that Opus 5 is going to require some amount of different engagement than either Fable 5 or Opus 4.8. Anthropic's Tariq explained a bit more about what was going on behind the scenes. In a post called The New Rules of Context Engineering for Claude 5 Models, Tariq explained that they had dramatically cut down the system prompts and built-in skills, which could explain some of the issues people have been having. Tariq wrote that Anthropic had removed 80% of the system prompt for Opus 5 and Fable 5 in Claude code. He said this resulted in zero change to their coding benchmarks, meaning basically that Anthropic found that they had been over-constraining Claude and potentially conflicting with user prompts and skills. In their internal work, they would often find traces where Claude was told to both leave documentation, but then on the other hand to not leave comments. Anthropic found that they were able to strip out a ton of these comments that were useful for earlier models and simply rely on surrounding context and judgment instead. Now, the upshot of this is that the rules of context engineering have completely changed, and a lot of skills will need to be rewritten. Tariq walked through a few of those rules that have changed, like using progressive context disclosure rather than frontloading everything, or no longer needing to use examples and instead being more descriptive. Now, this is a must-read if you're going to be engaging deeply with these new models. But the big takeaway around Opus 5 is that less is more. This generation of models simply don't need the same rigid frameworks as older models, which will have the benefit not only of better performance, but probably better efficiency as well. Tariq encouraged everyone to do a similar skills cleanup with this model release, and Anthropic have even rolled out a new command called Claude Doctor to help do that automatically. Now there were some much more positive takes on Opus 5 as well. YouTuber, entrepreneur, and developer Theo declared it a really good model. Now, he noted how confusing it is to have what seems to be a model that's both cheaper and better than Fable according to the benchmarks. And in his view, that framing stood up in practice, with Theo concluding, "This is probably the only model you need." One of the ways Theo tested the model was to make a plan to update his agentic coding platform to support Opus 5. The test was a bake-off with Opus 5 and Fable 5 writing completing plans. Theo immediately ran into a similar problem to the folks at Every, where Opus 5 couldn't use his existing skills. Instead telling him to take over and do some manual file management. Once that was resolved, each model reviewed and rated each other's plans as better. Both models preferred each other's plans, with Fable giving Opus a much higher ranking than Opus gave Fable. Getting a third opinion, GPT-5.6 Sol actually preferred Opus's plan to Fable's. Now, while Theo agreed that Opus isn't as intelligent as Fable or tenacious as 5.6, he did not believe that this left it without use cases. He pointed out that Opus is far more usage efficient than Fable for tasks where Anthropic models excel, Opus also isn't subject to Anthropic's data retention policies for Fable, so it can address a lot of use cases that involve sensitive data where Fable is a non-starter, i.e., pretty much every enterprise use case at this point. Interestingly, Theo thought another bonus was that the model just isn't that intelligent. And while that seems counterintuitive, one of Theo's gripes with 5.6 is that although it works until the problem is fixed, i.e., it is tenacious, in doing so, it writes in his estimation way too much code. That means that several weeks after release, Theo basically isn't merging any of the bloated code written by GPT-5.6. Theo found Opus, on the other hand, to be more diligent than Fable without resorting to the brute force of writing tons of code like GPT-5.6, representing, in his opinion, a pretty good balance at the frontier. He commented, I've been surprised. Opus sometimes is better than Fable. It often catches things Fable missed and has code that is more likely to actually work for the problems I want to solve than Fable does. I feel like I don't have to make that trade-off anymore. SOL would solve the problem at the cost of my sanity. Fable would make me feel great at the cost of the problem not being properly solved. Opus is the in-between, and I'm really liking it. Ben Davis on Theo's team agreed, but did also note some of the problems of Opus 5 stopping early, so in terms of interaction patterns, that may be one to watch for if you are starting to shift your behavior to Opus 5. Summing up a few different points of conversation, developer Kun Chen pointed out one, that the Opus 5 release really put a fine point on how useless benchmarks are in real life. Chen argued that, quote, Opus 5 is nowhere near Fable in practical use, not even close. Anyone who's used it meaningfully can tell this very quickly after a few tasks, yet Opus beats Fable on many benchmarks. Now, obviously, the experience of Theo and Claire Vaux maybe puts some comparison on that, but certainly it's more nuanced than a benchmark analysis would suggest. Chen also points out how pleasant it is to work with the model used to be a strength in Claude, but now it's not. He speculated, "It feels like both Anthropic and OpenAI are giving reinforcement learning from human feedback less care, in favor of scalable reinforcement learning that's machine verifiable. This almost looks like AI is directing humans to build a world that's more friendly for machines rather than humans. And most humans don't even realize they are being manipulated to help with that. Almost every new generation of frontier models now talk in more jargon, need more steering to do what you want, and are just less fun to work with." Now, a couple days on from the release, If you ask me right now for 10 people saying that the model sucked and 10 people saying that the model was good and another 10 people saying that the model both sucks and is good, I could find all of those things for you. But I think one of the really important points that is very easy to forget for those of us who are model omnivorous is that in the real world of average knowledge work, at least when it comes to your work environment, you're not sitting there choosing between Grok or OpenAI and Anthropic. You are locked into a specific company's models and those are your choices. And when you look at Opus 5 as a release in that context, not just as I think many of us terminally online folks on Twitter view it, in other words, as a replacement for Fable, as Fable gets restricted to the most expensive Anthropic plans, but instead as part of a complete model architecture for enterprise customers, it starts to make a little more sense. Arena's Peter Gostev wrote, Before this model, Anthropic was in a funny situation. They had a really exceptional model, made a lot of waves, but it was too expensive to use. You don't really want Fable running 24/7 and doing all sorts of things for you. Then the impression we got from Opus 4.8. It's a good model, but people weren't in love with it. Sonnet 5 didn't make much of a splash, and not many people wanted to switch to it. So Anthropic had a gap. They didn't have a really strong model that people really loved to use in that daily driver category. I think they have it now. It does look like a solid model. And so perhaps as we are judging how successful the model is going to be, the right question to ask is for enterprise users who are locked into the Claude ecosystem, does this represent a significant upgrade? And most, at least of the first analysis, is certainly yes compared to the Opus-48 model that for all intents and purposes was the main model that they were going to have access to. Now, for some, any model that's not state of the art just isn't going to make that big of a splash. And some are already looking forward to the future. AI commentator and news aggregator Andrew Curran wrote, "I think Fable 5.1 is ready, but Anthropic are saving it for OpenAI's next release. They will keep crossing swords like this from here on out. Soon it will make a lot more sense why Opus 5 was so performant and why the Fable class was preemptively moved to credits for most users." Chubby reposted that and said, "I fully agree with Andrew. I cannot imagine under any circumstances that Anthropic released Opus 5 with without already having a better model for the Fable tier in-house. The question, of course, is why it hasn't been released. Yet the answer is very simple if you open your eyes. The competition between OpenAI and Anthropic is fiercer than ever. GPT-5.6 was a resounding success, and Codex, with its current 10 million active professional users, is gaining increasing importance in a sector primarily dominated by Anthropic. Anthropic is holding off the launch of Fable 5.1 until OpenAI releases GPT-6, and that won't be long now. Axios reported that Sam Altman is briefing the White House on the new model next week, so it is essentially ready to launch. And yet for some, part of the reason that a major model could be released on a Friday and only really be splashy among insiders represents a bigger and yet inevitable trend. Arc Prize's François Chalet wrote, "The era of new model launches as big milestones will eventually come to an end. At some point they will simply be continuously updated with no widely publicized version number. Probably less than 2 years away." Now, I'm not totally sure about this. I think that it kind of depends on the capability unlock of each new model. And I certainly think the labs are going to have incentives to make each of them a big deal. But it is undeniably the case that as we move to these multi-model setups, or even increasingly routers that obfuscate the model behind an automated selector, the hugeness of these launches may be less of a big deal in the future. But I don't know. Let me know what you guys think. All I have to judge this is the comments and the download numbers. Anyways, guys, that is gonna do it for today's AI Daily Brief. Appreciate you listening or watching as always. And until next time, peace.