alt Hacker NewsDeepSeek-v4.1-Flash-1017
Comments
This model is why OpenAI and Anthropic want regulatory hurdles. We need more diversity like this, not less. A diverse ecosystem is the only long term solution to complex problems that I know of and the challenges that the AI world is presenting is no different.
The company's own news page: https://api-docs.deepseek.com/news/news261017
They are showing this new version scoring significantly higher across multiple coding benchmarks, beating Sol 5.6 (6.1 is the latest) across 10/11 benchmarks.
I am going back to DeepSeek for writing code. I was trying out using Sol and liking the result, but this should close a lot of that gap. I use Opus/Sol for an implementation plan and code review also use them for grilling. DeepSeek is also running verifications.
so i did some apples to bananas comparison.
i let 4 different agents review the same dirty code base (200+ files changed, c++, game engine):
- deepseek v4.1 flash via deepseek API: took 20m to find one p1 issue that was relevant and a p2 issue that was irrelevant
- glm 5.3 flash NVFP4 via TP2 DGX Sparks: took 1h 5m to find a p3 documentation issue
- gpt 5.6 sol: took 5m30s for one p1 and one p2 issue. both relevant
- gpt 6.1 sol: took 4m30s for two p1 and one p2 issue. both relevant.
Worth to mention: none of those findings overlapped.
I let deepseek v4.1 flash and gpt 6.1 sol both plan fixes for the summary of the findings. in this particular scenario: gpt 6.1 sol planned to increase a buffer size from 128 to 256 to fit the 191 elements that blew past a test scenario. it included some headroom. deepseek v4.1 flash on the other hand decided to make it exactly 191 to satisfy the test, for example.
The TPS decode on deepseek is impressive. I never had a second / feeling of being left aone and though "damn this thing is working". Despite all my efforts, it took ~4x longer in a review, found less and planned worse than GPT 6.1 Sol. Wall clock time. I don't have token counts for comparison. If I were to grade those models from 1-10 (10 = top tier GG), I'd give deepseek a solid 6.5 and 6.1 sol a 7.5-8.0 maybe.
Hilarious sentence: "DeepSeek-V4.1-Flash-1017 is a minor update to DeepSeek-V4.1-Flash, delivering substantial improvements across reasoning, coding, agentic tasks, and visual benchmarks." (emphasis mine)
Its a great update, I ran a benchmark comparing verbosity and intelligence compared to old weights.
- Reasoning effort = 1 is now actually usable and produces 2.5x less tokens than old weights, without intelligence lost.
- Overall 20% less tokens at each reasoning effort.
- Same intelligence and same tool-calling pattern.
More details in blog post https://sference.com/blog/deepseek-v41f-1017-benchmark
I've been using DS v4.1 Flash daily for nearly a month now on Agent Swarm experiments. I've tested it against GLM-5.3-flash and Opus 5.5. DS 4.1 comes out tops for me. Opus 5.5 is still the smartest, but DS 4.1's speed and price more than make up for the "intelligence" gap. I've ran Chief/Builder/Reviewer swarms autonomously for multiple days. I think my new workflow is going to have DS 4.1 Flash do all the engineering and just have Opus review its work. 4.1 Flash really is a very impressive model. Kudos to the Deepseek team.
I really like the having DeepSeek Flash + Opus 5.5 as advisor/reviewer at the moment. I run through a $200 Claude Code subscription in less than 24 hours now just for a single project, and I hate fiddling with different accounts every day. Now that I get a 200 USD api credits inlcuded with the subscriptions I have rewritten my coding agent workflow completely.
Mario one-shotted by deepseek-v4.1-flsh-1017 (in high mode, not even max) with sound and programmatically generated assets in under 5 minutes: https://mario-dsv4-1-flash-1017.pagey.site/
Honestly given how good the open weight models are getting I don’t see a viable business model for OpenAI and Anthropic anymore. The profit is in the compute, not the model. OpenAI and Anthropic spent all their money renting compute from others so yes they can sell lightweight model inference but they have a massive structural disadvantage now as others can always sell good-enough inference much cheaper.
There’s also zero stickiness to models and most companies now have their own internal routing utilities so devs and users can just keep the same API endpoint and move between models with a single line of code swap.
Unless something changes with the big labs and fast, they look pretty f*cked.
It passed my threshold for not caring to try new models back with DS4, I use it for literally everything. Announcements of new models in the series always feel curious, because it doesn't really feel like there's much left to fix that would improve on what's already there for any task I know of.
It's a bit like obsessing over patch releases of some software after a major upgrade already solved the original problem long ago.
A negative datapoint - We have been evaluated prior Deepseek Flash models (both the original 4.1 flash and the smaller 4.0 flash) and they have been overthinkers. Our workload is well defined but moderately difficult coding tasks and among open weight models GLM5.3 flash performs best with Qwen3.8 flash next a close second. DeepSeek 4.1 flash fails at lower thinking levels and spends a huge amount of tokens (an order of magnitude more) at the highest thinking levels where it is able to successfully complete the tasks. I will evaluate this newest release since they specifically claim "the model reduces task execution time and output length", but keeping my hopes low.
I have started phasing into DeepSeek v4 (Dwarfstar quant from Antirez) as my daily driver and it's going very well. It can one shot iOS apps on my Mac Studio, it's a little slow and sometimes needs retries but can definitely do medium difficulty real work for me.
I didn't try the last flash model because I would need to use SSD streaming (I have 128 GB), but the benches for this release are so compelling (and it using less tokens to get tasks done) it might be worth the throughput hit. I will probably give it a try when I have some downtime.
It feels very freeing to be doing useful work on a local model on my own hardware with my own coding harness. The TCO argument is starting to look compelling too.
Above GLM 5.3, nice! That is a very capable model. I do wish that we will see some smaller "flash" models again though - running this one locally is tough (very costly).
Still need 2 or 3 days to properly quant this. But it is really good model to run if you have 64GiB or larger RAM on Mac. On M5 Ultra, you get ~3400 tok/s for prompt processing and ~50 tok/s for token generation, quite usable state.
Here are some pelicans. The "max" one took over 11 minutes: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...
Definitely an improvement on deepseek-v4-flash-0731: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...
I essentially only use DeepSeek. Almost always the flash variant. I've found that I don't really need frontier models if I'm doing most of the thinking myself and just using the agent to output code quickly.
Getting a lot of errors from OpenRouter's endpoint "Failed to get response from the AI model; retried 5 times (total retry wait time: 60.71 seconds) Last error: 429 Provider return error"
It works when it works, but will probably be flaky AF for a few days.
Works well with Copilot:
COPILOT_PROVIDER_API_KEY=<OR key> \
COPILOT_PROVIDER_BASE_URL=https://openrouter.ai/api/v1 \
COPILOT_MODEL=deepseek/deepseek-v4.1-flash-1017 \
COPILOT_PROVIDER_MAX_PROMPT_TOKENS=1048576 \
COPILOT_PROVIDER_MAX_OUTPUT_TOKENS=700000 \
COPILOT_OFFLINE=true \
copilot
I haven't put this into test as much yet.
As a Luna6 proponent, this has a potential as it looks slightly more intelligent and at the similar price point.
I can't wait for a smaller moe dinstillation from this excellent model
Thank you deepseek, we are eating good today.
I am a huge DeepSeek fan, but it cannot come close to Opus 5.5 on UI design. A $20 Claude subscription that I use some months and cancel the others along with Deepseek API is great.
Once read, I think from G Huntley, "pick a model and stick with it," they have been making that very easy, daily driver since V4.
Looks very impressive especially considering the price and speed. Let's wait for AA to confirm how well their benchmarks hold up.
At this point I am way more excited by DeepSeek releases than anything from OpenAI or Anthropic.
The speed and cost/intelligence ratio is just outrageous.
The native web search api is 10x cheaper than OpenAIs. I hope they will release the web search api as a standalone API.
Have security companies considered that these open models could be used as sleeper agents even when they're deployed outside of China? Is that a serious attack vector or more like science fiction?
Those are crazy numbers for a Flash model. Hopefully they can scale what they are doing to 4.1 Pro.
This is a great model. I know we are all looking for the next best thing, but I'd be pretty happy running this for awhile.
By far my favorite model.
At the beginning of the month I realized I had more tokens and more capabilities on Opencode at 8$/month via DS 4.1 flash, than I had on a 200$ account in June with the american providers.
Literally both more tokens and capabilities than Opus 4.8/GPT 5.5.
Even if you pay per token, you're still unlikely to outspend american providers.
I use Opus 5.5 and swarms of about 50 deepseek models to reverse engineer old videos games. The scale is roughly 600k loc generated from pure asm, simply by knowing what the old compiler is. Truly staggering stuff.
What would be the best local harness to run this? It feels there are 1000s now.
Here's a small benchmark that compares the new checkout with the previous one:
Another big win for opensource AI.
BTW who would have thought China would be the champion of freedom and opensource in the AI arena, while US is fighting for corporate/government control.
The new model is amazing. Had it build a full Angular Material & CDK components showcase via DeepSeek Harness and it easily did the best job of any recent models I've seen.
Does DeepSeek provide the inference code along with the weights? I read here recently that for models where only the weights are released, providers have to implement which could be different than what the labs themselves use.
I wonder how you would be able to reliably rely on a model like that for agentic work, even for implementation only after careful planning, how are you making sure you are not wasting tokens reviewing and correcting it over and over again ?
This and (the not quite yet open) Kimi K2.8 are my main coding engines, and I find them incredibly cost efficient. I find Deepseek in particular is really good with Svelte work.
Right now I use each of them in their own respective coding harnesses, deepseek-harness and kimi-code, but I do wonder what they are like inside something like opencode.
Very keen to try this out tonight. What coding harness do people like to use with these models?
Why no Jinja-format chat template?
I'd never heard of the BabyVision benchmark they do so well on. Pretty interesting stuff:
My favorite model just got better. Gotta love what deepseek is doing
they must've been using this since 2-3 days when it became free again in OpenCode, it's noticeable smarter, beats GLM
beating K3 and GLM-5.3 is kinda insane
And Christmas comes early this year.
DeepSeek v4.1 Flash was a beast.
It was already cheaper and more powerful than Haiku 5.5 which was released like a full month after it... by a trillion dollar american "frontier" company.
Can't wait to see what this one has in store.
"a minor update" they say before showing off benchmarks that display a substantial improvement.
DS isn't better than Sol and Gemini in the benchmarks, but after working with them all for a while now I'm beginning to understand why people find DS _feels_ so much better.
First, it's the tok/s and TTFT from providers. Deepseek blazes through work - at quite good quality - and you don't find yourself waiting for your agents to finish.
Second is the openness. Being able to see the CoT and not being randomly blocked for cyber review. You're never wondering if it's stuck or if it will randomly fail. Combined with the great cost value.. My next subscription will probably be Z.ai