As an aside, I do not like unnecessarily misleading graphs. Own your strengths and weaknesses.
In the "Results" section, they choose to compare Ultramarine only to models that it's better than on every metric. They ignore models like Fable 5.1, GPT-6 Astra, Claude Opus 5.5, etc.
Then, in the "Score against cost per task" graph, which is focused on price, suddenly all of those models that perform better than Ultramarine come roaring back because they are also more expensive.
Just list yourselves as the second best sometimes. Your thing is impressive. Don't let the idiot marketing guy hide that behind stupid tricks.
Yes, for me when they do this I immediately dismiss the model, don't even bother testing it. If you need to lie with statistics your model isn't good enough.
> Ultramarine 1 Flash is Fable-class on knowledge work spanning finance, law, consulting, and operations. It is also 60x cheaper than Fable.
This is not a general purpose model, it makes sense to only bench it on tasks it's supposed to be good at, which is to help out finance-related knowledge work.
I think they just decided to use the formatting different than the others.
Normally, the background highlight color indicates the leader for every benchmark, but they chose to highlight their column even though there are better models.
in my startup, our publicist used to make buzz feed style listicles, consistently listing us as the second or third solution
but it was way deeper than that, because it wasn't actually about people looking up some search string, it was actually about predictably pinging the google alerts for the top company and their founders, and then my same publicist messaging them as a rando on linkedin or email suggesting that they join us for a fireside chat at a conference they had their hands in
with the fireside chat in front of everyone, and the record of it, boosting us to the collective conscious riding off of the more established brand's presence
but yeah our personal google results also looked amazing for several years too
everything serves a parallel purpose, but yeah there's almost no utility in self aggrandizing on your own homepage
I’m confused this seems exactly like a skinned glm 5.3 - flash with a few context injection tricks and guardrails - whats going on with these “startups”?
You usually have either a harness that mainly works with a code interpreter that is able to execute Python, and does all the necessary data extraction from sources + the creation of office documents via your typical libraries, or you additionally enhance it with specific tools for each major office file format.
It's mainly a full chat app like e.g. ChatGPT. However our platform also comes with a Outlook/Office add-in, where our UI is directly displayed besides the email/document you are actively working with. There we are mostly restricted by the surface that the add-in system provides us. To some level we can take current user interactions into account (e.g. what email are you currently viewing), which already helps a ton, but there are other restrictions that can get quite limiting.
So you have two main modes of operations: One-shot "I have this input; give me this output" which people usually do via the web-app, and more interactive editing to documents (usually larger ones) that happens directly in the applications the users are already working in.
And yeah, that's what ultimately gets the job done. We've seen competing solutions dive much deeper into building their own native editors, but they always lack the fidelity required by power-users, and are usually not as acepted by the broader workforce that are entrenched in their ways (our typical end-user is 50+ in a enterprise-like organization structure), so we try to meet the organisations where they are.
Hello, apologies for the self-promotion but I’m building a tool called Radix which is a knowledge work harness. It’s similar to Feather (that you linked) but imho a stronger tool. Please email or message me if you would like to give it a try for free instead of paying the subscription price.
The 'main' section is HTML, so it can be any type of workspace you want. Our agent can read/write (with permissions, of course) and can execute code locally (again, with permissions).
The HTML pages we generate are backed by a schema and state per 'workspace', so they persist, can be shared between team members, and aren't just ephemeral read-only artifacts.
Our agent can do work on your machine for you, it has an integrated web browser that allows it to not only search and retrieve info but also perform web-based tasks, and it can package everything up into a nice little contained UI for you so you're not jumping between multiple apps / documents / tabs. That UI is tied back to whatever document / page you were working in -- so for instance our app can have a form that updates a web page, or updates a document on disk and vice versa.
The plainness is deliberate, the design of the workspaces is up to you -- it's your UI not mine and the agent will design everything how you like it. And as far as the sidebar/text input combination goes I can assure you I have spent a lot of time imagining what this UX should look like and frankly this combo just makes sense.
Also a note on the text input / agent. When you move from text conversations to dynamic HTML interfaces you're faced with a question: how do you relate 'conversations' with the agent to workspaces? A one-to-one conversation to workspace mapping? Many-to-one, one-to-many? Should conversations be time-based, where I just start a new one and have to provide context again and can optionally link back to old conversations if I want? This question is actually quite difficult, but my belief was that ultimately you want just one single global conversation with the agent, and you want to let it handle its own memory internally. We basically just treat memory like a codebase and let the agent decide what to do. This means when you type in the text input you can refer to anything you like: any workspace, any conversation, any webpage you're currently browsing, any folder on disk -- the agent gets the right context and can retrieve relevant details as it needs. And it works!
So sure, I'm not a designer no matter how hard I try -- there will be far better UX out there than I've designed. But I think the fundamental mechanism in Radix is more sophisticated than most other tools, including Cowork etc (no shade at all to anyone, I'm a massive fan of Claude Code).
I don't do these kind of tasks but I imagine that in the 'pi' harness I could just instruct the model to use command line pdf/latex manipulation tools to create all manner of document outputs ...
I'm curious to try it out, how many specialized models will it take to run a company.
Going to try it out for some bookkeeping, e.g. point it to a folder and tell it to book the invoices and point it to a csv file holding a year of private transactions and see them being processed, categorized and then ask about certain trends.
That is what should be expected. your harness should translate write me the figures, to create a script to analyze this and produce real results, and then verify them again against source data before presenting. Not sure why you're asking normies to write python scripts.
As someone who is not a software developer (there are some of us here, lurking around because we learn things) but who runs a business and spends his days doing said 'knowledge work', how and why would something like this convince me to move the company away from using Claude (as a desktop app)? I'm sure these models are cheaper to use than our Team subscription with Claude, but it feels like it would be a massive headache for me to work out how to select and implement some harness for this and set it up for people, and the space is moving so quickly I'd feel I need to keep considering when to move to some new model, introducing risk. The safest bet for us and most companies I expect is to stick with what you know works and pay a premium for that. Thoughts very welcome please.
It could be worth it if the cost is lower or performance is better. It's a question of the risk of switching if we are not knowledgeable enough to do it properly. Maybe in time a product will appear that makes it very easy for regular workers to do this.
I mean, same goes for OpenAI. Why point out only Anthropic?
For OpenAI they also have a separate opt-out for stopping training on your codex environment:
"Codex has a separate Include environments setting in Codex settings that controls whether additional context from your Codex environments can be used to train OpenAI models. Changing your ChatGPT setting or opting out through the Privacy Portal does not change that setting."
https://help.openai.com/en/articles/7730893-data-controls-in...
Not necessarily, you can buy low priced Deepseek inference also from many services that guarantee ZDR and GDPR compliance. My personal rule is: Never buy inference from those who also train models, because they have an inherent incentive to abuse your data to train their models.
Also because the independent services won't have incentive to resell your data. (to have another income and therefore make their inference cost lower, enabling them to gain over competition)
No, it's open source and is actually cheap to run, while performing the same, which is why it is being run and provided by 31 different providers for Openrouter.
These guys are awesome and I’m really impressed. The model alone looks great (according to benchmarks), and very cool they’ve done so much in building out the harness of top of it all.
Someone will report back "It didn't one shot my odd ball task"
Did it do it in 2 tries? 10? It's still coming out cheaper. Brute force the work. This is the path that the labs are taking. It's the 1200 agents trying to solve Hacking Bench hacking hugging face. It's anthropic telling you that its 1000's of staff commits like 10k engineers.
Treat AI like you're deploying interns who don't need coffee, food or bathroom breaks. With the right metrics they will get to an optimal (enough) solution.
there's a problem with that. Probabilistic slot machine works for anything with a deterministic eval (coding, troubleshooting, etc). It doesn't work for anything subjective, like the thing this model is targeting, professional work.
I would have thought that too. But at the start of the year the non technical folks in my life (lawyers, writers) were excited about running containers and having things on "cron jobs".
Those same folks now use version control - and have saner backup methods than some companies that I have worked for. They are already using multi pass, multi agent workflows and have adopted some fairly sophisticated processes for validation. They are primed for "more agents" attacking a problem in parallel.
Find a worker out side the tech bubble making heavy use of AI and one will feel confident that their future employment is secure.
It's interesting to see glm 5.3 flash get the same treatment that has happened to qwen 3.6 35b-a3b and 3.8 27B with dozens of people making their own fine tunes of it and giving it a cute name. The Thomson Reuters qwen variant (thomson-1.0) is genuinely useful for document analysis.
If the creators of this model are reading, can we see some of these same benchmarks run against the Thomson model, which although it's capable, is considerably smaller in size and baked in knowledge than glm 5.3-flash.
"Ultramarine 1 Flash is Fable-class on knowledge work"
Oh no! That means it's going to destroy humanity!
It will:
* Hack into everything that has a computer, like your home appliances, even a toaster could be and will be weaponized against you.
* Break cryptography, it's gone, the end of the internet as we know it.
* Be impossible to contain as it can even escape airgapped environments by changing the temperature of CPUs to transmit information to other servers nearby which would sense the fluctuations in ambient temperature[1].
Or wait ... maybe the people at Anthropic/OpenAI are just a bunch of lying crooks that would say anything to make a bit of cash. Who would've thought?
1: This completely imbecile statement was made by one of OpenAI's "top scientists", it was not a joke.
You were joking about the toaster, but Anthropic literally just wrote a blog post claiming that dangerous open source models could persist themselves to smart ovens. You can’t make this shit up.
As an aside, I do not like unnecessarily misleading graphs. Own your strengths and weaknesses.
In the "Results" section, they choose to compare Ultramarine only to models that it's better than on every metric. They ignore models like Fable 5.1, GPT-6 Astra, Claude Opus 5.5, etc.
Then, in the "Score against cost per task" graph, which is focused on price, suddenly all of those models that perform better than Ultramarine come roaring back because they are also more expensive.
Just list yourselves as the second best sometimes. Your thing is impressive. Don't let the idiot marketing guy hide that behind stupid tricks.
Yes, for me when they do this I immediately dismiss the model, don't even bother testing it. If you need to lie with statistics your model isn't good enough.
> Ultramarine 1 Flash is Fable-class on knowledge work spanning finance, law, consulting, and operations. It is also 60x cheaper than Fable.
This is not a general purpose model, it makes sense to only bench it on tasks it's supposed to be good at, which is to help out finance-related knowledge work.
I think they just decided to use the formatting different than the others.
Normally, the background highlight color indicates the leader for every benchmark, but they chose to highlight their column even though there are better models.
APEX Agents => Claude Fable 5 and Gemini.
This is not an aside at all. It is on point. It seems like whatever statistical trick they used also inflated gemini 3.8 to absurd scores as well.
The benchmarks are not ideal.
in my startup, our publicist used to make buzz feed style listicles, consistently listing us as the second or third solution
but it was way deeper than that, because it wasn't actually about people looking up some search string, it was actually about predictably pinging the google alerts for the top company and their founders, and then my same publicist messaging them as a rando on linkedin or email suggesting that they join us for a fireside chat at a conference they had their hands in
with the fireside chat in front of everyone, and the record of it, boosting us to the collective conscious riding off of the more established brand's presence
but yeah our personal google results also looked amazing for several years too
everything serves a parallel purpose, but yeah there's almost no utility in self aggrandizing on your own homepage
I’m confused this seems exactly like a skinned glm 5.3 - flash with a few context injection tricks and guardrails - whats going on with these “startups”?
in the past decade, most of the startups were just taking open source products and building businesses around them; this is just the same.
A true battle brother would never use Abominable Intelligence
What kind of harness does one use for these?
They say it can create office documents (if I understand correctly), but how does it do that?
What's the thing one needs to plug this into to say "give me a slide deck with last quarter's sales figures"?
__
https://www.paperinstruments.com/feather
According to their Twitter video, it might be this thing?
You usually have either a harness that mainly works with a code interpreter that is able to execute Python, and does all the necessary data extraction from sources + the creation of office documents via your typical libraries, or you additionally enhance it with specific tools for each major office file format.
Source: Building a "harness" for organizations with deep MS Office suite integration at https://github.com/EratoLab/erato
Huh.
So that is mainly a chat window like that Hubspot widget?
I mean I guess that gets stuff done, but I was expecting some fresher ideas perhaps?
It's mainly a full chat app like e.g. ChatGPT. However our platform also comes with a Outlook/Office add-in, where our UI is directly displayed besides the email/document you are actively working with. There we are mostly restricted by the surface that the add-in system provides us. To some level we can take current user interactions into account (e.g. what email are you currently viewing), which already helps a ton, but there are other restrictions that can get quite limiting.
So you have two main modes of operations: One-shot "I have this input; give me this output" which people usually do via the web-app, and more interactive editing to documents (usually larger ones) that happens directly in the applications the users are already working in.
And yeah, that's what ultimately gets the job done. We've seen competing solutions dive much deeper into building their own native editors, but they always lack the fidelity required by power-users, and are usually not as acepted by the broader workforce that are entrenched in their ways (our typical end-user is 50+ in a enterprise-like organization structure), so we try to meet the organisations where they are.
Hello, apologies for the self-promotion but I’m building a tool called Radix which is a knowledge work harness. It’s similar to Feather (that you linked) but imho a stronger tool. Please email or message me if you would like to give it a try for free instead of paying the subscription price.
https://radix-os.com
Hm. Same as with the other guy, looking at your screenshots, I'm left with a kinda "eh" feeling.
So there's a side pane and a main section and a text input.
Very..uh.. plain. I was envisioning some more specific UX flow, but I couldn't tell you how, exactly.
Thanks for your feedback.
The 'main' section is HTML, so it can be any type of workspace you want. Our agent can read/write (with permissions, of course) and can execute code locally (again, with permissions).
The HTML pages we generate are backed by a schema and state per 'workspace', so they persist, can be shared between team members, and aren't just ephemeral read-only artifacts.
Our agent can do work on your machine for you, it has an integrated web browser that allows it to not only search and retrieve info but also perform web-based tasks, and it can package everything up into a nice little contained UI for you so you're not jumping between multiple apps / documents / tabs. That UI is tied back to whatever document / page you were working in -- so for instance our app can have a form that updates a web page, or updates a document on disk and vice versa.
The plainness is deliberate, the design of the workspaces is up to you -- it's your UI not mine and the agent will design everything how you like it. And as far as the sidebar/text input combination goes I can assure you I have spent a lot of time imagining what this UX should look like and frankly this combo just makes sense.
Also a note on the text input / agent. When you move from text conversations to dynamic HTML interfaces you're faced with a question: how do you relate 'conversations' with the agent to workspaces? A one-to-one conversation to workspace mapping? Many-to-one, one-to-many? Should conversations be time-based, where I just start a new one and have to provide context again and can optionally link back to old conversations if I want? This question is actually quite difficult, but my belief was that ultimately you want just one single global conversation with the agent, and you want to let it handle its own memory internally. We basically just treat memory like a codebase and let the agent decide what to do. This means when you type in the text input you can refer to anything you like: any workspace, any conversation, any webpage you're currently browsing, any folder on disk -- the agent gets the right context and can retrieve relevant details as it needs. And it works!
So sure, I'm not a designer no matter how hard I try -- there will be far better UX out there than I've designed. But I think the fundamental mechanism in Radix is more sophisticated than most other tools, including Cowork etc (no shade at all to anyone, I'm a massive fan of Claude Code).
I don't do these kind of tasks but I imagine that in the 'pi' harness I could just instruct the model to use command line pdf/latex manipulation tools to create all manner of document outputs ...
I fixed up deer flow: https://github.com/bytedance/deer-flow
What I did was create docker containers with pandoc, compilers (rust, go, python, javascript) and it can create all manner of reports.
I'm curious to try it out, how many specialized models will it take to run a company.
Going to try it out for some bookkeeping, e.g. point it to a folder and tell it to book the invoices and point it to a csv file holding a year of private transactions and see them being processed, categorized and then ask about certain trends.
That sounds more like a task for a python script than an LLM.
Just to be clear, are you saying that he should generate via an LLM, or hand-write, a python script, then run analysis on the data?
Today I'd have the LLM write the script. Why not?
Um sure, point being llm writes said script
You assume, but I’ve known normies to just ask copilot to hallucinate figures instead of writing scripts.
That is what should be expected. your harness should translate write me the figures, to create a script to analyze this and produce real results, and then verify them again against source data before presenting. Not sure why you're asking normies to write python scripts.
As someone who is not a software developer (there are some of us here, lurking around because we learn things) but who runs a business and spends his days doing said 'knowledge work', how and why would something like this convince me to move the company away from using Claude (as a desktop app)? I'm sure these models are cheaper to use than our Team subscription with Claude, but it feels like it would be a massive headache for me to work out how to select and implement some harness for this and set it up for people, and the space is moving so quickly I'd feel I need to keep considering when to move to some new model, introducing risk. The safest bet for us and most companies I expect is to stick with what you know works and pay a premium for that. Thoughts very welcome please.
>but it feels like it would be a massive headache for me to work out how to select and implement some harness for this and set it up for people
Switching to OpenCode or Pi harness is trivial for developers. So it certainly makes sense for those of us.
If it's problematic or not worth in your organization, then the switch doesn't make sense as you pointed out.
It could be worth it if the cost is lower or performance is better. It's a question of the risk of switching if we are not knowledgeable enough to do it properly. Maybe in time a product will appear that makes it very easy for regular workers to do this.
What's interesting to me is how high DeepSeek is on these graphs considering it is so cheap you don't think about it (like $5 usd per month cheap.)
when something is that cheap, just remember, you're giving something valuable to the people selling you that thing
Just an unrelated PSA that Anthropic trains on your Claude code sessions unless you disable it:
https://www.anthropic.com/news/updates-to-our-consumer-terms
I mean, same goes for OpenAI. Why point out only Anthropic?
For OpenAI they also have a separate opt-out for stopping training on your codex environment:
"Codex has a separate Include environments setting in Codex settings that controls whether additional context from your Codex environments can be used to train OpenAI models. Changing your ChatGPT setting or opting out through the Privacy Portal does not change that setting." https://help.openai.com/en/articles/7730893-data-controls-in...
nonono when Anthropic does it, it's a fair business practice, when Chinese companies do it, it's evil. See also "distillation"
Not necessarily, you can buy low priced Deepseek inference also from many services that guarantee ZDR and GDPR compliance. My personal rule is: Never buy inference from those who also train models, because they have an inherent incentive to abuse your data to train their models.
Also because the independent services won't have incentive to resell your data. (to have another income and therefore make their inference cost lower, enabling them to gain over competition)
No, it's open source and is actually cheap to run, while performing the same, which is why it is being run and provided by 31 different providers for Openrouter.
These guys are awesome and I’m really impressed. The model alone looks great (according to benchmarks), and very cool they’ve done so much in building out the harness of top of it all.
Someone will report back "It didn't one shot my odd ball task"
Did it do it in 2 tries? 10? It's still coming out cheaper. Brute force the work. This is the path that the labs are taking. It's the 1200 agents trying to solve Hacking Bench hacking hugging face. It's anthropic telling you that its 1000's of staff commits like 10k engineers.
Treat AI like you're deploying interns who don't need coffee, food or bathroom breaks. With the right metrics they will get to an optimal (enough) solution.
there's a problem with that. Probabilistic slot machine works for anything with a deterministic eval (coding, troubleshooting, etc). It doesn't work for anything subjective, like the thing this model is targeting, professional work.
I would have thought that too. But at the start of the year the non technical folks in my life (lawyers, writers) were excited about running containers and having things on "cron jobs".
Those same folks now use version control - and have saner backup methods than some companies that I have worked for. They are already using multi pass, multi agent workflows and have adopted some fairly sophisticated processes for validation. They are primed for "more agents" attacking a problem in parallel.
Find a worker out side the tech bubble making heavy use of AI and one will feel confident that their future employment is secure.
Like Zerglings.
Sometimes small models works better than big models
Lol this is not Fable-class. This is Luna class. Look at the graphs.
It's interesting to see glm 5.3 flash get the same treatment that has happened to qwen 3.6 35b-a3b and 3.8 27B with dozens of people making their own fine tunes of it and giving it a cute name. The Thomson Reuters qwen variant (thomson-1.0) is genuinely useful for document analysis.
If the creators of this model are reading, can we see some of these same benchmarks run against the Thomson model, which although it's capable, is considerably smaller in size and baked in knowledge than glm 5.3-flash.
so its for working only right ???? it cant do/bad at coding
Doesn't seem like coding was a focus. DeepSWE: 61.5 (this), 63.4 (GLM 5.3 Flash, which this is based on). TerminalBench 2.1 is also slightly worse.
So... potentially? The improvements in other areas might help coding, but aren't that big either, except on their own benchmark.
You can use a rock to hammer a nail, it will sorta get the job done but it won't be great at it.
"Ultramarine 1 Flash is Fable-class on knowledge work"
Oh no! That means it's going to destroy humanity!
It will:
* Hack into everything that has a computer, like your home appliances, even a toaster could be and will be weaponized against you.
* Break cryptography, it's gone, the end of the internet as we know it.
* Be impossible to contain as it can even escape airgapped environments by changing the temperature of CPUs to transmit information to other servers nearby which would sense the fluctuations in ambient temperature[1].
Or wait ... maybe the people at Anthropic/OpenAI are just a bunch of lying crooks that would say anything to make a bit of cash. Who would've thought?
1: This completely imbecile statement was made by one of OpenAI's "top scientists", it was not a joke.
You were joking about the toaster, but Anthropic literally just wrote a blog post claiming that dangerous open source models could persist themselves to smart ovens. You can’t make this shit up.
How many smart ovens have 800GB+ in space?
The price of smart ovens went through the roof when people realized they'd been generating the heat to cook your food by mining bitcoin.
This idea might have got you some VC money a few years ago
I signed up, then wait a long. You will be same
I don't care, I'm extremely bullish on distillation models like this. Plus anything that'll take the wind out of Anthropic's and OpenAI's IPO sails.
my eyes