[1]: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/...
[2]: https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system...
Why do you think your conception of the dangers are more accurate than all the scientists who have spent their lives studying this?
I know everybody wants the tell all story of the clever ideas that were developed over the last ~3 years at Anthropic and OpenAI, but what I really want to thumb through is DeepSeek's notebook of "brilliant but didn't quite make the cut" ideas.
They must be trying some truely bonkers stuff to be able to land this much architecture novelty in their full releases.
The bad news is that the original v4 flash was 284B, which was large but still somewhat reasonable for running locally. This one is 552B so almost twice that, so the huge gains in benchmark scores make sense - it's not really flash anymore, imo.
I've no idea about actual performance vs benchmaxxing, though deepseek was fairly trustworthy as far as Chinese models go. If that holds (and if it doesn't think forever, as deepseek 4 sometimes did) it's probably the newest king of the hill amongst open weights models.
It does include vision, and they do something funky with KV cache so it's very efficient: "[...] these designs reduce the global KV cache footprint to 890 bytes per token — roughly 1/4 of DeepSeek-V4-Flash". I do appreciate the high focus on efficiency, but at this point we sure could use a flash-flash version.
@edit: I couldn't make sense what the actual parameter count is, with the addition of Engram memory. To my understanding the 4.1 flash is 552B parameters you want in vram or ram, out of which ~16B is active (8B for prefill). It also includes additional 196B Engram memory which you can put on an SSD. I think.
Assuming that's correct 256 GB memory is insufficient to even load the model at q4 - you'd be 1GB short, assuming you can fill it to 100% (so no mac). You'd also want some for kv cache of course. A 256 GB desktop with some extra VRAM from GPU could run it, but normal consumer boards get real slow once you fill 4 slots so you'll probably want quad channel which is Threadripper or above territory.
It uses fewer active parameters, though. (8B or 14B instead of always 13B)
So ... flash indeed.
Original Flash good fit for dual Spark / Strix Halo machines. This one would require third party quants and even then 4 machines.
Edit: Most of added weights/size are Engrams?
> Overall, DeepSeek-V4.1-Flash has 552B backbone parameters and 196B Engram parameters, activating 8B parameters per token during prefill and 16B during decode.
Those can stay on SSD. So I guess / it possible, that non-engram portion is still FP4 of ~same size! Need to read tech report.
552B in ~FP4, 306GB.
196B of FP8 Engrams, another 204GB, not necessary to keep in RAM.
KV cache sees another 4x size reduction, just 900MB for 1M.
can't wait for deepseek v4.1 pro
I suffix everything with "Reply in English", and even so I‘m getting lots of Chinese.
It also seems to be more willing to just do whatever you ask of it. My favourite benchmark for this is to ask it to download a rom for an old game, that I own. Legal in my juristiction but the US models (except Grok) have a tendency to refuse it.
Even easier: just have them review a large codebase of yours that accidentally has a OOB access bug. Even with no consequences and even if the codebase is truly yours you get blocked.
And of course "find vulnerabilities in..." prompts are out of the question, whereas Chinese models happily oblige.
Every model release seems like it packed with wonderful research and advancements.
Underlying it all is that any architecture can be trained to the same convergence just difference in compute utilization both in training and inference
> Causal Encoder-Decoder (CED) architecture: a 40-layer Transformer organized as a 20-layer causal encoder followed by a 20-layer decoder. With CED, the decoder's global KV cache is projected from the final encoder hidden states rather than derived from each decoder layer's own hidden states. This allows the model to activate only 8B parameters per token during prefill and 16B during decode, substantially improving cost efficiency for input-heavy agentic workloads.
> these designs reduce the global KV cache footprint to 890 bytes per token — roughly 1/4 of DeepSeek-V4-Flash.
Faster prefill, lower kv cache (~1GB / 1m context is insane).
> The model supports a continuously controllable reasoning effort setting (integer 1–100) that trades inference cost for accuracy.
Benchmarks are benchmarks, to be seen if they translate to real-world use, but they seem to have focused a lot on post-training with "agentic" scores looking good. "world knowledge" is obviously lower than higher param models.
I was hoping for a bit more, but it's still 100% faster for a very good price, so I won't complain.
In Pi (pi.dev), it tells me it's definitely Claude by Anthropic, via the API via curl it tells me it's "probably ChatGPT", its very funny.
“DeepSeek launching v4.1 flash cheaper and more capable than v4 pro”
399 points | 19 hours ago | 216 comments
[1] https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/...