It's so refreshing to see DeepSeek's tech report[1] full of juicy details; meanwhile, something like Fable's system card[2] is like 70% "safety", 10% "model welfare" to make sure little Claude isn't distressed, and 20% benchmark numbers.
As I also said on Twitter - it really amazes me how fearless Deepseek are. Every single model release is packed with new and crazy clever ideas and somehow, they always commit to training them at near frontier scale.
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.
rao-v
I'm surprised more people aren't talking about the cache hit price: $0.003 per million tokens. I have a feeling that the price of 1 million tokens transmitted over the internet is more expensive than cache hit. Are we close to making the chat completion API obsolete because the cost of context transfer over network is going to dominate the task total cost?
Here's the same token usage priced at different rates: a real long-running coding task, medium codebase, 447 turns.
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.
revolvingthrow
I think it's very clear that DeepSeek is obviously the best AI lab in the world.
Every model release seems like it packed with wonderful research and advancements.
impulser_
Initial impressions: this is a really strong model and the fact that they reduced prices at the same time makes it an awesome backup model to use when your primary subscription runs out and you need to bridge a few days before it resets.
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.
I ran it on all seven reasoning levels supported by OpenRouter, but the reasoning token counts suggest to me that it doesn't actually support seven different levels. This is one of my biggest problems with OpenRouter - their abstraction layer makes reasoning levels harder to reason about.
reasoning_level reasoning_tokens
none 0
minimal 6,520
low 11,873
medium 5,678
high 9,779
xhigh 10,197
max 13,386
Deepseek v4.1 flash is able to solve it some of the time. I've found it burns through more reasoning tokens than any other model. Google models like Gemini 3.8 Flash are still dominating and is able to one-shot most evals while being the cheapest.
I'm curious what other unique evals people are running.
Should be the link ( now that it works again! :) )
mentalgear
OpenCode Go is currently running a 4x usage promo on DeepSeek v4.1 flash, not a bad way to get your feet wet (even if their cache hit prices are probably still very sub-optimal)
comments (10)
[1]: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/...
[2]: https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system...
kouteiheika
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.
rao-v
Here's the same token usage priced at different rates: a real long-running coding task, medium codebase, 447 turns.
GPT-6-astra DeepSeek v4.1 Flash, $0.003 cache hit DeepSeek v4.1 Flash, $0.006 cache hit Hypothetical: same DeepSeek input/output rates, but cache priced so it accounts for 66% of the bill. This cache it improvement makes the model x2-x2.5 more efficient on a long horizon tasks in terms of cost.k9294
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.
revolvingthrow
Every model release seems like it packed with wonderful research and advancements.
impulser_
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.
LaurensBER
I ran it on all seven reasoning levels supported by OpenRouter, but the reasoning token counts suggest to me that it doesn't actually support seven different levels. This is one of my biggest problems with OpenRouter - their abstraction layer makes reasoning levels harder to reason about.
Update: explained here: https://api-docs.deepseek.com/guides/thinking_mode/That says it supports three levels - low, high, max, and maps them out like this:
(But it looks like "none" is a valid option too.)simonw
Deepseek v4.1 flash is able to solve it some of the time. I've found it burns through more reasoning tokens than any other model. Google models like Gemini 3.8 Flash are still dominating and is able to one-shot most evals while being the cheapest.
I'm curious what other unique evals people are running.
pampas
Should be the link ( now that it works again! :) )
mentalgear
swiftcoder