Always glad to see more open-weight models, but this caption on the 2nd demo image had me do a double-take: "Land or Water Generalization Experiment: We recreated the viral X puzzle by asking Beam to create a fixed 180×90 grid for longitudes -179° to 179° and latitudes -89° to 89°, with 16,200 points. This puzzle is a few days old, so could not appear in the training data, thus testing the model’s generalization. Beam gets 95.5% coverage right, putting us between Opus 5 (92.5%) and Fable 5 (97.8%), which shows how well it generalizes to novel new tasks."
Oof, no, this "puzzle is a few days old" is incorrect even if it's a social media trend just recently. Asking a model to generate a world map in this way is _at least_ from August 2025 as it appeared on LessWrong at that time: https://www.lesswrong.com/posts/xwdRzJxyqFqgXTWbH/how-does-a...
Yeah I remember when the original post about this came out. Def not recent. Though I think their point survives in that they didn't exactly RL on this.
Model weights (what is being tested here) don't inherently "access the web" when inference is running. If the model has access to a web search tool, that's a different story.
> Beam is a sparse Mixture-of-Experts model with 501 billion total parameters, 23 billion active, built for coding, reasoning, and agentic workloads.
> Beam’s capabilities come from major investments in both pretraining and reinforcement learning (RL). We pretrained the model on 23.8 trillion diverse, curated, high-quality tokens from the web and proprietary licensed datasets, matching or outperforming available similar-sized open base models. In parallel, we developed the algorithms, training environments, and infrastructure needed to sustain high-compute RL at exceptional scale. Our high-compute RL run generated over 100 million rollouts on 10.5K NVIDIA GB300 GPUs over 4 weeks of training.
Early access, no weights no tech details, just a sign up here for info
I'm all for more open models, but talk is cheap and this is a rather pointless announcement without anything backing it up. Publish your weights and HF repo or shut up IMO.
Is that all that a company about to give away the product of 10,000 GPUs running for a month gets to be now? give it away without a single promotional post or shut up? I support open source as much as the person but this is pretty caustic.
And also a "proprietary data set" hahaha... Probably just means they don't want to show it, and it is data, that either they shouldn't have, or that there is nothing special about their training data and it is just meant to sound like there is some secret ingredient, while there is none.
Not sharing the data is pretty standard because 1) it tends to get the lawyers involved and 2) good data is critical for getting good results.
Imo you can get better results with great data and generic modeling techniques than with incredible modeling techniques and crappy data. Because if you have crappy data, you won’t even know if your model is good because your evals will also be bad.
This is why Anthropic is throwing a fit about the Chinese distillation “attacks”. Clean reasoning traces are gold.
Companies pay lots of money for proprietary agentic trajectories which are used during RL. These are things like "Task: summarize stock levels for months end accounting" which then traces the task though using SAP to look at different SKU stock levels, exporting them and generating summary Excel spreadsheets.
This is very different to the "scrape the internet" datasets that a table stakes for training a LLM.
Data has copyright issues, so one can't share it generally without getting permissions from all of the copyright holders. The data is not theirs to share, anyways. The derived (learned) weights are a different matter.
this is very normal for frontier lab companies. you need good data either synthetic or labelled (all the chinese open source models have their own armies of data labelers)
What kind of company or organization is Reflection?
I think it is ever more important to realize who is releasing models rather than what the models do and how they compare.
Because models iterate at breakneck speed, looking at today's benchmarks is only useful for someone using the models today. Whereas if one builds a product on top of it, or commits to one for a project or team, the company or organization behind it, is far more important. Will they exist in a few months? Do they need a business-model? Are they subsidizing usage with venture capital and how long can they keep this up?
Looks like general reasoning is their target. As for who:
> The startup was launched in March 2024 by Misha Laskin, who led reward modeling for DeepMind’s Gemini project, and Ioannis Antonoglou, who co-created AlphaGo, the AI system that famously beat the world champion in the board game Go in 2016.
with the obligatory:
> Investors in Reflection AI’s latest round include Nvidia, Disruptive, DST, 1789, B Capital, Lightspeed, GIC, Eric Yuan, Eric Schmidt, Citi, Sequoia, CRV, and others.
Bigger and still worse than existing free Chinese models that are smaller? Open weight models are nice, but at this point it seems western models are very far behind Chinese ones, despite Chinese companies publishing a lot of their findings. I hope we get more open models and more providers, as being stuck with a model from China or US with no competition is risky.
Google does do a great job with Gemma models. It's one of the few language models actually good at language. OpenAI's top closed models can't even write norwegian correctly.
It takes time / few iterations to get it right (and it's moving target), but yes, expensive trial, my personal feeling is that they went a bit too high, at the same time who knows, maybe good move – as they're saying RL didn't plateau. It feels like they had something like $100M budget for it?
I hear this so much from westerners who have no connection to the people or the country.
-China passed a law forbidding companies from replacing workers with AI.
-China regularly punishes CEOs and corrupt government officials who do bad things.
China promotes open source to the world enabling everyone in every country to have equitable access.
From living in china for over 20 years and talking to many many people, the majority is happy with the trajectory of their country.
Could any for these be said of the west?
I won't get into specifics on which countries regularly throw bombs on children, use starvation and blockades as weapons, and completely disregard the massive dissatisfaction of their citizens...but it sure isn't china.
Please clean your own house first before you complain about your neighbor.
This sentiment was really common circa 2005, or at least general anti-China rhetoric, where I lived in the Tri-state. Funnily enough, I have never felt more a cog, living here in the U.S., than I do right now. Maybe you are right, but my guess is wherever you live, yours is a bit of a glass house as well.
Yeah, we should only promote, like... Finnish, Norwegian, maybe Swiss products, like Apertus. Not this unethically-made baby oil[1] from repressive torment nexuses of USA or China.
I feel like this is a marketing miss. If they had held their announcement until the model was released, I would have grabbed it and started running it through my benchmarks. It probably doesn’t get a place in the rotation based on their own description of its performance, but now the weights live on the server, I’m probably following them on HF and I will remember to check in every time I ls the models folder. With the announcement only, none of that happens and I’m likely to forget about this by the time it actually gets released.
The email harvest move just doesn’t fit where we/they are in the cycle. There are established players and a buffet of models to choose from (plus a ton of empty hype). The first move at this point for any new entrant should be to show, not tell. Even an API only release with the promise to open weight would be better (actually probably all around better since most people can’t run this locally).
I wish this lab and all the labs releasing the best. It’s a brutal landscape to sink millions of dollars into for a guaranteed “behind x model from a year ago” evaluation. But, I do believe there is genuine innovation left to uncover.
It's pretty clear from their framing ("Beam advances the Western open-weight frontier") that one of their main selling points is not being a Chinese lab.
I can't imagine that mattering to many individuals, but I guess someone out there has a government contract that forbids the use of foreign models
Multiple independent approaches are cool and all but fully open source model training (datasets, pipeline, checkpoints) should be taking advantage of being open and share runs/budget between different entities.
I disagree. Sure let them play and see if they can improve. But this model has more compute and more training data than the predecessors it fails to surpass. That only means their training regime is inferior if their predecessors did so much more with so much less. That inferiority should not be encouraged.
Reflection is explicitly marketed as the 'US' DeepSeek
seems like they are aiming to provide both inference and RLaaS for american companies and western govts. even if they never fully beat deepseek if they get close enough the fact that they're American will help them close deals
Is it worse than the top open-weight Chinese models? Yes, it is, but at least the West has joined the party, and hopefully they will iterate on this and keep up the pace. The Chinese labs will certainly release new and powerful versions soon, so it's all about relative pace right now.
I remember being in the room with pretraining day 1 to help monitor the training job launch. Watching this model train from day 1 has been an amazing experience!
Outside of ML metrics, you're monitoring the health of every piece of hardware in the system. You need to make sure that you have every GPU, every CPU, the PCIe buses, the networking fabric are all working without any errors. You need to ensure that you can respond as fast as possible to any possible error. One bad component can bottleneck the entire job.
I don't find open-weight models that impressive anymore. MiMo-V2.6 already showed that you can have a not so crazy architecture and enough compute, the bottleneck is then just the data. OAI and Anthropic are largely the frontier models because of the synthetic data they made. They have a large customer base and have the user's data as well as knowing what tasks their customers use the models for and what domain they should get synthetic data for.
Interesting. Have heard about the need to create synthetic data for LLM's, but didn't realize it was such an important factor. Although, how big a factor synthetic data is the next question, but guess there is no way to truly verify how much difference it makes with these closed models.
This is literally what the labs have been doing over the past year. The whole idea of emergence is a lie, there is some interpolation and superhuman long evaluations the models can do, but almost all the gains are from synthetic data. They hire thousands of professionals and pay them as much as 200$/hour to create many tasks that they want the model to perform and use these to teach the model on how to do it with RL. OAI had 30,000 contractors from Mercor for Sol 5.6.
When you see Opus suddenly becoming great at blender or some other 3D graphics, that's because they hired professionals and had them do similar tasks that people are looking for. They keep having better professionals at each iteration and so the quality improves. There is no emergence or "General" intelligence. The model doesn't learn to become better at a task because of scaling laws or emergence or whatever they might wanna say, it is literally RL on tasks that they want the model to perform well on.
Does this one also routes to Claude under the hood like Reflection 70B did? I recall they even run a basic regex to remove "Claude" from the output. Then they promised to be completely transparent on the postmortem (they claim they had no idea what had happened), but the postmortem never came
Beggar choosing: my kingdom for more 90B-133B MoE local models. Especially with disk offload, that is a function/performance sweet spot for Mac workstations with 64GB-128GB of RAM.
Releasing open-weight models at this scale is a massive milestone for the community. Access to transparent model internals is foundational for trustworthy systems.
Instead of yet another mediocre but fully-made-in-the-West open model (alongside Mistral, Trinity, Poolside, Inkling, etc etc) I'd really love for a Western neloab start the same way Qwen did: by focusing on post-training. Qwen's first release was a Llama 1 finetune [1]! Once they made it useful, they started working their way back in the stack to also do their own pretraining, etc. Starting with pretraining feels like such a waste: there's millions of dollars of crystallized compute and data sitting around in the Chinese model weights. Why not start with one of those, and only work your way back to pretraining once you've released something you can prove is useful?
At least for RL, you don't need a base model — the rollouts are run in an inference engine with an instruction-tuned model using a chat template. You can start with just that!
I am really curious why after Qwen-3.8-flash and deepseek-v4.1-flash newer open weight models (or proprietary but they don't tell) don't use n-grams. It looks (to me) like they are a cheep way to add more knowledge to the model.
What do I mean by cheap? You can rely on the SSD to retrieve the relevant tokens as no computation is needed meaning you can leverage storage (or cpu ram if you don't have unified memory) to serve part of the model which (to my understanding) is much cheaper to get than GPU RAM.
Anyone know what am I missing? Or is it that the pace of iteration for labs slow enough that they can't actually leverage it yet?
I think they'll become more popular in the coming months. Also Gemma 4 PLE (April) is similar to DeepSeek Engram in a lot of ways, just with 1-grams.
On the proprietary model point: I'm personally curious about whether heavy n-gram offload is one reason Anthropic keep driving down their token vocabulary size (the other reason being eliminating the LM head gradient bottleneck).
Hey since it's an open-weight model, I would love to know: how much model safety alignment have you done? Did you do any sort of post-training and what restrictions are in place
Ideally i would like to place my own restrictions and align from scratch, currently I am resolved to do harness alignment using tools like Prismor but would love to do my own post training alignment
the performance chart puts the better open source models behind the fold making it seem like it outperforms them... but it doesn't! all for open source models but this announcement is misleading
Any time a new lab shows up, folks complain about how their models are worse. Really? It would be nice if a new comer comes from no where and beats everyone, but that's rarely the case. The good thing is that other labs/people are figuring out how to build this, and if they keep at it then this is as bad as it gets for them and it would hopefully get better. A new entrant to the market is good for everyone.
honest question: how honest do you think people are about their improvements and performance compared to objective results when all you do is praise them?
I'm on the waiting list... Couldn't find any download option, so I suppose it is only obtainable through their API. Strange way to distribute open weights model.
It's true that no benchmark communicates the whole picture, and we won't really know how it behaves until weights are out, but the performance here doesn't seem particularly groundbreaking just based on the benchmark.
The meaningful test begins after the weights arrive: fixed harness, network disabled, repeated trials, and real latency, memory, energy and cost per completed task.
Very little in terms of the layers they use. Calling it now, they are using global layers everywhere, making the model basically unusable due to high KV Cache use.
Nitpicking but I really wish this benchmarks table were easier to read. Should show which columns win in each row and should not require horizontal scrolling to see across.
Open model that is not yet open or widely accessible via API. Primarily comparing to non-SOTA models like Inkling and GLM 5.2. Included comparison to GLM 5.3 and DeepSeek V4.1 Flash in the table, but not in the charts (I assume they would make them look bad). Also no results from AA Index or Arena.
very curious to see more about what kinds of hardware you can run this on and the perf. characteristics… on the face of it, it seems like optimizing for inference speed might(?) be good for running on smaller hardware, but i suppose it could be the other way around and it is actually much resource-hungrier for the number of parameters, etc. …
ok so they are comparing themselves to and claim to be beating GLM's last generation GLM 5.2 model, GLM 5.3 Flash is a monster, this is honestly embarassing
If you don't buy into "America good, China bad" narrative, this new entrant & release by Inclusion Ai is a lot more exciting by every measurable metric.
I am a great fan of open weights model, but off late I am starting to loose track on the capabilities of the latest models released and now a days most of the open weights models are above 100 B params which is not going to run in our laptops. What happened to Jev hype? A 501B Beam model is not going to help with it (Non AR Schema driven responding under 1 second).
I am more interested with a SOTA Frontier 8B-10B model. Is this even possible?
All of those are good references. Other folks in the thread are missing distinctions between pre training (~the internet + curated sources) and post training (~instructions and RL)
If you ask a model, they will generally tell you where to get data. Modern frontier models have the large advantage of having tens if not hundreds of millions of users providing use cases to train against to improve their responses.
Note that this sort of distillation is NOT for pre-training data (which is tens of trillions of tokens). I think the allegations against Chinese companies by Anthropic is more so that they distill SFT data (which is good for post-training, but you still need a strong base model)
pretty wild that, per public sources, Reflection AI raised over $4 billion and hit a $25 billion valuation while operating in total stealth, without ever releasing a single public product until now. Beam (501B) seems be their first-ever model drop. Or am I missing something?
Always glad to see more open-weight models, but this caption on the 2nd demo image had me do a double-take: "Land or Water Generalization Experiment: We recreated the viral X puzzle by asking Beam to create a fixed 180×90 grid for longitudes -179° to 179° and latitudes -89° to 89°, with 16,200 points. This puzzle is a few days old, so could not appear in the training data, thus testing the model’s generalization. Beam gets 95.5% coverage right, putting us between Opus 5 (92.5%) and Fable 5 (97.8%), which shows how well it generalizes to novel new tasks."
Oof, no, this "puzzle is a few days old" is incorrect even if it's a social media trend just recently. Asking a model to generate a world map in this way is _at least_ from August 2025 as it appeared on LessWrong at that time: https://www.lesswrong.com/posts/xwdRzJxyqFqgXTWbH/how-does-a...
Yeah I remember when the original post about this came out. Def not recent. Though I think their point survives in that they didn't exactly RL on this.
It could be old but still not be part of the training set.
Yes, but then you can’t use “it’s 2 days old” to assert that it can’t be in the training set.
Seems sloppy.
I'd love to see these tests repeated for the current frontier models...
Seems rather sloppy to not validate it wasn’t in their training or RL data.
I don't think age of the puzzle even matters, all models have search capacities these days
> all models have search capacities these days
one would hope that they disable websearch and internet access (maybe all tools?) when doing generalization testing?
Model weights (what is being tested here) don't inherently "access the web" when inference is running. If the model has access to a web search tool, that's a different story.
Is search part of the model or the harness?
3 replies →
If they made that statement and knowingly had search enabled, it would essentially be fraudulent.
> Beam is a sparse Mixture-of-Experts model with 501 billion total parameters, 23 billion active, built for coding, reasoning, and agentic workloads.
> Beam’s capabilities come from major investments in both pretraining and reinforcement learning (RL). We pretrained the model on 23.8 trillion diverse, curated, high-quality tokens from the web and proprietary licensed datasets, matching or outperforming available similar-sized open base models. In parallel, we developed the algorithms, training environments, and infrastructure needed to sustain high-compute RL at exceptional scale. Our high-compute RL run generated over 100 million rollouts on 10.5K NVIDIA GB300 GPUs over 4 weeks of training.
Early access, no weights no tech details, just a sign up here for info
I'm all for more open models, but talk is cheap and this is a rather pointless announcement without anything backing it up. Publish your weights and HF repo or shut up IMO.
Is that all that a company about to give away the product of 10,000 GPUs running for a month gets to be now? give it away without a single promotional post or shut up? I support open source as much as the person but this is pretty caustic.
13 replies →
They claim that they will release the model as open weights later this month.
That means that they have the 31th of October as the deadline to make true their claims.
The fact that they give early access to some may mean that they want some beta testers before the public release.
And also a "proprietary data set" hahaha... Probably just means they don't want to show it, and it is data, that either they shouldn't have, or that there is nothing special about their training data and it is just meant to sound like there is some secret ingredient, while there is none.
Not sharing the data is pretty standard because 1) it tends to get the lawyers involved and 2) good data is critical for getting good results.
Imo you can get better results with great data and generic modeling techniques than with incredible modeling techniques and crappy data. Because if you have crappy data, you won’t even know if your model is good because your evals will also be bad.
This is why Anthropic is throwing a fit about the Chinese distillation “attacks”. Clean reasoning traces are gold.
This isn't true.
Companies pay lots of money for proprietary agentic trajectories which are used during RL. These are things like "Task: summarize stock levels for months end accounting" which then traces the task though using SAP to look at different SKU stock levels, exporting them and generating summary Excel spreadsheets.
This is very different to the "scrape the internet" datasets that a table stakes for training a LLM.
Xiaomi released a fairly developer-centric dataset like this here: https://huggingface.co/datasets/XiaomiMiMo/MiMo-V2.6-RL-oss
SpreadsheetRL is another fairly specialized dataset: https://spreadsheet-rl.github.io/
Data has copyright issues, so one can't share it generally without getting permissions from all of the copyright holders. The data is not theirs to share, anyways. The derived (learned) weights are a different matter.
2 replies →
this is very normal for frontier lab companies. you need good data either synthetic or labelled (all the chinese open source models have their own armies of data labelers)
> Beam is undergoing final red-teaming and evaluations. You can sign up here for early access to the model.
> We will release the weights, technical report, model card, and developer artifacts later this month.
What kind of company or organization is Reflection?
I think it is ever more important to realize who is releasing models rather than what the models do and how they compare.
Because models iterate at breakneck speed, looking at today's benchmarks is only useful for someone using the models today. Whereas if one builds a product on top of it, or commits to one for a project or team, the company or organization behind it, is far more important. Will they exist in a few months? Do they need a business-model? Are they subsidizing usage with venture capital and how long can they keep this up?
Is reflection a company? University lab? NGO?
Looks like general reasoning is their target. As for who:
> The startup was launched in March 2024 by Misha Laskin, who led reward modeling for DeepMind’s Gemini project, and Ioannis Antonoglou, who co-created AlphaGo, the AI system that famously beat the world champion in the board game Go in 2016.
with the obligatory:
> Investors in Reflection AI’s latest round include Nvidia, Disruptive, DST, 1789, B Capital, Lightspeed, GIC, Eric Yuan, Eric Schmidt, Citi, Sequoia, CRV, and others.
https://techcrunch.com/2025/10/09/reflection-raises-2b-to-be...
Can't open that website with an ad blocker.
1 reply →
[dead]
I thought it would be interesting to look at some key figures vs another contemporary model in the same weight class (DeepSeek V4.1 Flash)
At first blush the benchmarks are impressive, but to paraphrase Linus: "Talk is cheap, show me the weights." :-)
yeah agreed, let's normalize "show me the weights" in this era
Bigger and still worse than existing free Chinese models that are smaller? Open weight models are nice, but at this point it seems western models are very far behind Chinese ones, despite Chinese companies publishing a lot of their findings. I hope we get more open models and more providers, as being stuck with a model from China or US with no competition is risky.
Google does do a great job with Gemma models. It's one of the few language models actually good at language. OpenAI's top closed models can't even write norwegian correctly.
It takes time / few iterations to get it right (and it's moving target), but yes, expensive trial, my personal feeling is that they went a bit too high, at the same time who knows, maybe good move – as they're saying RL didn't plateau. It feels like they had something like $100M budget for it?
[flagged]
I hear this so much from westerners who have no connection to the people or the country. -China passed a law forbidding companies from replacing workers with AI. -China regularly punishes CEOs and corrupt government officials who do bad things. China promotes open source to the world enabling everyone in every country to have equitable access. From living in china for over 20 years and talking to many many people, the majority is happy with the trajectory of their country. Could any for these be said of the west? I won't get into specifics on which countries regularly throw bombs on children, use starvation and blockades as weapons, and completely disregard the massive dissatisfaction of their citizens...but it sure isn't china. Please clean your own house first before you complain about your neighbor.
4 replies →
This sentiment was really common circa 2005, or at least general anti-China rhetoric, where I lived in the Tri-state. Funnily enough, I have never felt more a cog, living here in the U.S., than I do right now. Maybe you are right, but my guess is wherever you live, yours is a bit of a glass house as well.
1 reply →
Yeah, we should only promote, like... Finnish, Norwegian, maybe Swiss products, like Apertus. Not this unethically-made baby oil[1] from repressive torment nexuses of USA or China.
[1] https://i.imgur.com/rJSG019.jpeg
I feel like this is a marketing miss. If they had held their announcement until the model was released, I would have grabbed it and started running it through my benchmarks. It probably doesn’t get a place in the rotation based on their own description of its performance, but now the weights live on the server, I’m probably following them on HF and I will remember to check in every time I ls the models folder. With the announcement only, none of that happens and I’m likely to forget about this by the time it actually gets released.
The email harvest move just doesn’t fit where we/they are in the cycle. There are established players and a buffet of models to choose from (plus a ton of empty hype). The first move at this point for any new entrant should be to show, not tell. Even an API only release with the promise to open weight would be better (actually probably all around better since most people can’t run this locally).
I wish this lab and all the labs releasing the best. It’s a brutal landscape to sink millions of dollars into for a guaranteed “behind x model from a year ago” evaluation. But, I do believe there is genuine innovation left to uncover.
This appears to be larger than DeepSeek v4.1 Flash, more expensive to run, and worse on every measured metric.
Am I missing something?
> Am I missing something?
It's pretty clear from their framing ("Beam advances the Western open-weight frontier") that one of their main selling points is not being a Chinese lab.
I can't imagine that mattering to many individuals, but I guess someone out there has a government contract that forbids the use of foreign models
Reflection raised on the idea of creating the "American Deepseek Project"
12 replies →
Multiple independent approaches are cool and all but fully open source model training (datasets, pipeline, checkpoints) should be taking advantage of being open and share runs/budget between different entities.
We're still at the stage where every new entrant is welcome in my opinion. Doesn't need to be record-breaking upon initial release.
It depends! If a startup is entering with a large model to face other larger models, it must be better at least in 1 meaningful dimension.
500B params performing worse than other OSS of the same size is pretty meaningless if no one will use it.
2 replies →
I disagree. Sure let them play and see if they can improve. But this model has more compute and more training data than the predecessors it fails to surpass. That only means their training regime is inferior if their predecessors did so much more with so much less. That inferiority should not be encouraged.
9 replies →
Reflection is explicitly marketed as the 'US' DeepSeek
seems like they are aiming to provide both inference and RLaaS for american companies and western govts. even if they never fully beat deepseek if they get close enough the fact that they're American will help them close deals
deepseek is from the evil east, this is from the virtuous west
Apparently there is more to making good models than copying everything on the internet.
12 yards long, 2 lanes wide, 65 tons of American Pride! Canyonero! Canyonero!
Yes it's (hopefully) not distilled from every single major American provider.
new entrant in this weight class, US lab.
Beam goes brrrr
Is it worse than the top open-weight Chinese models? Yes, it is, but at least the West has joined the party, and hopefully they will iterate on this and keep up the pace. The Chinese labs will certainly release new and powerful versions soon, so it's all about relative pace right now.
I remember being in the room with pretraining day 1 to help monitor the training job launch. Watching this model train from day 1 has been an amazing experience!
What sort of outputs or telemetry is monitored on a large pre-training run?
Outside of ML metrics, you're monitoring the health of every piece of hardware in the system. You need to make sure that you have every GPU, every CPU, the PCIe buses, the networking fabric are all working without any errors. You need to ensure that you can respond as fast as possible to any possible error. One bad component can bottleneck the entire job.
2 replies →
I don't find open-weight models that impressive anymore. MiMo-V2.6 already showed that you can have a not so crazy architecture and enough compute, the bottleneck is then just the data. OAI and Anthropic are largely the frontier models because of the synthetic data they made. They have a large customer base and have the user's data as well as knowing what tasks their customers use the models for and what domain they should get synthetic data for.
Interesting. Have heard about the need to create synthetic data for LLM's, but didn't realize it was such an important factor. Although, how big a factor synthetic data is the next question, but guess there is no way to truly verify how much difference it makes with these closed models.
This is literally what the labs have been doing over the past year. The whole idea of emergence is a lie, there is some interpolation and superhuman long evaluations the models can do, but almost all the gains are from synthetic data. They hire thousands of professionals and pay them as much as 200$/hour to create many tasks that they want the model to perform and use these to teach the model on how to do it with RL. OAI had 30,000 contractors from Mercor for Sol 5.6. When you see Opus suddenly becoming great at blender or some other 3D graphics, that's because they hired professionals and had them do similar tasks that people are looking for. They keep having better professionals at each iteration and so the quality improves. There is no emergence or "General" intelligence. The model doesn't learn to become better at a task because of scaling laws or emergence or whatever they might wanna say, it is literally RL on tasks that they want the model to perform well on.
4 replies →
Does this one also routes to Claude under the hood like Reflection 70B did? I recall they even run a basic regex to remove "Claude" from the output. Then they promised to be completely transparent on the postmortem (they claim they had no idea what had happened), but the postmortem never came
Reflection AI is a completely different company with no relation (as far as I can tell) to the model you are referencing from 2024
If that’s the case i take it all back. I thought this was another venture by Matt Shummer’s Reflection Ai
Beggar choosing: my kingdom for more 90B-133B MoE local models. Especially with disk offload, that is a function/performance sweet spot for Mac workstations with 64GB-128GB of RAM.
Releasing open-weight models at this scale is a massive milestone for the community. Access to transparent model internals is foundational for trustworthy systems.
Instead of yet another mediocre but fully-made-in-the-West open model (alongside Mistral, Trinity, Poolside, Inkling, etc etc) I'd really love for a Western neloab start the same way Qwen did: by focusing on post-training. Qwen's first release was a Llama 1 finetune [1]! Once they made it useful, they started working their way back in the stack to also do their own pretraining, etc. Starting with pretraining feels like such a waste: there's millions of dollars of crystallized compute and data sitting around in the Chinese model weights. Why not start with one of those, and only work your way back to pretraining once you've released something you can prove is useful?
1: https://en.wikipedia.org/wiki/Qwen
I think there aren't any recent base models to post train on. Labs don't release them any more.
At least for RL, you don't need a base model — the rollouts are run in an inference engine with an instruction-tuned model using a chat template. You can start with just that!
1 reply →
https://huggingface.co/IFM
This is point of Nemotron series
> Where frontier open models like Kimi K3 remain ahead on raw capability, Beam's advantage is efficiency at inference time.
It's great to see a company that acknowledges it still needs improvement instead of making false claims.
I am really curious why after Qwen-3.8-flash and deepseek-v4.1-flash newer open weight models (or proprietary but they don't tell) don't use n-grams. It looks (to me) like they are a cheep way to add more knowledge to the model.
What do I mean by cheap? You can rely on the SSD to retrieve the relevant tokens as no computation is needed meaning you can leverage storage (or cpu ram if you don't have unified memory) to serve part of the model which (to my understanding) is much cheaper to get than GPU RAM.
Anyone know what am I missing? Or is it that the pace of iteration for labs slow enough that they can't actually leverage it yet?
DeepSeek Engram paper published: 12th January
Qwen3.8 Flash Next release date: 26th August
DeepSeek V4.1 Flash release date: 10th September
Current date: 6th October
I think they'll become more popular in the coming months. Also Gemma 4 PLE (April) is similar to DeepSeek Engram in a lot of ways, just with 1-grams.
On the proprietary model point: I'm personally curious about whether heavy n-gram offload is one reason Anthropic keep driving down their token vocabulary size (the other reason being eliminating the LM head gradient bottleneck).
Hey since it's an open-weight model, I would love to know: how much model safety alignment have you done? Did you do any sort of post-training and what restrictions are in place
Ideally i would like to place my own restrictions and align from scratch, currently I am resolved to do harness alignment using tools like Prismor but would love to do my own post training alignment
the performance chart puts the better open source models behind the fold making it seem like it outperforms them... but it doesn't! all for open source models but this announcement is misleading
Yes. All the link made me realize is that I should checkout Deepseek 4.1 flash
What kind of machine does one need to run this model? What's the usual setup someone would have for running this?
Any time a new lab shows up, folks complain about how their models are worse. Really? It would be nice if a new comer comes from no where and beats everyone, but that's rarely the case. The good thing is that other labs/people are figuring out how to build this, and if they keep at it then this is as bad as it gets for them and it would hopefully get better. A new entrant to the market is good for everyone.
Reflection isn't really "from no where", they have huge financial backing, 10K of the latest GPUs, and many ex leads from the established labs.
honest question: how honest do you think people are about their improvements and performance compared to objective results when all you do is praise them?
participation awards are not helpful.
Someone should name their next model "Workhorse" just for SEO reasons.
It's interesting how the industry converged to this very term, given that very less work is being done by horses since quite a while.
I guess future AI agents might market things as a "workman" for the same reason, despite less work being done by people :)
Thankfully there wasn't mistagging, we could have ended with workjackass.
I'm on the waiting list... Couldn't find any download option, so I suppose it is only obtainable through their API. Strange way to distribute open weights model.
I'm a big fan of open-weight models.
It's true that no benchmark communicates the whole picture, and we won't really know how it behaves until weights are out, but the performance here doesn't seem particularly groundbreaking just based on the benchmark.
The meaningful test begins after the weights arrive: fixed harness, network disabled, repeated trials, and real latency, memory, energy and cost per completed task.
[flagged]
Very little in terms of the layers they use. Calling it now, they are using global layers everywhere, making the model basically unusable due to high KV Cache use.
If Beam "rivals GLM 5.2 on reasoning" does it mean it's as good as GLM 5.3 Flash? (a much smaller model)
Open-weight at 501B is huge. What was the eval setup - was web search disabled to test generalization?
Nitpicking but I really wish this benchmarks table were easier to read. Should show which columns win in each row and should not require horizontal scrolling to see across.
That was exciting... Will come back later when the weights are out and gguf'd.
Access is currently limited. We'll contact you if early access becomes available.
Minimum hardware to run this model?? doable in consumer hardware?
2x 128GB unified memory machines with RDMA could run this at Q4. Not very fast.
The price of that kit is rapidly approaching $10k.
[flagged]
This is like an ad for how great Deepseek V4.1 Flash is.
Open model that is not yet open or widely accessible via API. Primarily comparing to non-SOTA models like Inkling and GLM 5.2. Included comparison to GLM 5.3 and DeepSeek V4.1 Flash in the table, but not in the charts (I assume they would make them look bad). Also no results from AA Index or Arena.
very curious to see more about what kinds of hardware you can run this on and the perf. characteristics… on the face of it, it seems like optimizing for inference speed might(?) be good for running on smaller hardware, but i suppose it could be the other way around and it is actually much resource-hungrier for the number of parameters, etc. …
Noticed some good questions made dead at the bottom of this thread. Odd...
ok so they are comparing themselves to and claim to be beating GLM's last generation GLM 5.2 model, GLM 5.3 Flash is a monster, this is honestly embarassing
Am I the only one who thought of the BEAM VM after the first word of thee title?
nopes
If you don't buy into "America good, China bad" narrative, this new entrant & release by Inclusion Ai is a lot more exciting by every measurable metric.
https://github.com/inclusionAI/Ling
There are many measurable metrics and I don't see any that seem impressive. Care to share the ones you found exciting?
That seems to be from ~1y ago. What am I missing?
You can directly check the link to Huggingface in their readme; they are constantly updating it on HF.
BTW, this is also an AI lab from China
I don’t buy into that narrative, but I don’t understand why Ling’s release is more exciting?
Congrats to the reflection team
I am a great fan of open weights model, but off late I am starting to loose track on the capabilities of the latest models released and now a days most of the open weights models are above 100 B params which is not going to run in our laptops. What happened to Jev hype? A 501B Beam model is not going to help with it (Non AR Schema driven responding under 1 second).
I am more interested with a SOTA Frontier 8B-10B model. Is this even possible?
Suppose I inherited a data center spanning several hundred acres full of GPUs and free electricity.
Where do I get the data?
I mean, this many models. They have to start somewhere.
I guess public datasets on HuggingFace and some shadow libraries content is enough to start.
e.g. fineweb dataset is 50TB https://huggingface.co/datasets/HuggingFaceFW/fineweb
There are a lot of open-research on pre-training, post-training and RL data mixtures and sourcing.
I recommend checking papers from Datalogy, Nvidia Nemotron, Ai2 (Ollmo, Tulu, ...) and the recent model from Aleph Alpha if you want to learn more.
All of those are good references. Other folks in the thread are missing distinctions between pre training (~the internet + curated sources) and post training (~instructions and RL)
I heard you should ask Claude about this. Preferably with thousands of accounts, routed through residential proxies
If you ask a model, they will generally tell you where to get data. Modern frontier models have the large advantage of having tens if not hundreds of millions of users providing use cases to train against to improve their responses.
https://scale.com/data-engine - you just buy it.
any prices anywhere for anything specific?
1 reply →
You can also hire teams to create data for you for higher quality.
Get data from Claude. That's what the Chinese (allegedly) do.
Note that this sort of distillation is NOT for pre-training data (which is tens of trillions of tokens). I think the allegations against Chinese companies by Anthropic is more so that they distill SFT data (which is good for post-training, but you still need a strong base model)
forget the data....sell it and go live your life!
fauxpen like OpenAI? says open but no weights. Feels like a bid to get a buyout before the weights go public. Doesn't even have technical details.
[flagged]
[flagged]
[flagged]
[flagged]
[dead]
[dead]
pretty wild that, per public sources, Reflection AI raised over $4 billion and hit a $25 billion valuation while operating in total stealth, without ever releasing a single public product until now. Beam (501B) seems be their first-ever model drop. Or am I missing something?
First model from them, but I think they released some sort of agent last year? https://sequoiacap.com/article/reflection-ai-asimov