DeepSeek V4

“It will promise to keep it in English from here on out, then switch back to Chinese after two or three exchanges.”

“the LLM generated SSE and AVX implementations of the compression and decompression algorithms that almost doubled performance with a single core”

“both would do things like change the response format from one message to another in the same chat”

“I honestly don't know of any OSS open weight models from the US labs as good as Kimi K3 or Deepseek V4 though.”

“Other sensitive questions (eg: Tiananmen Square) it clams up unless you ask the question a specific way”

“DeepSeek has a phenomenal cacheing architecture that specifically can work with less-than-ideal clients like this and get cache hits anyway.”

“Deepseek is one of the worst in terms of hallucination rate according to artificial analysis' benchmark”

“DeepSeek V4 hallucinates like crazy and often forgets explicitly mentioned parts of the context.”

“I'm also experimenting with DeepSeek 4 and it seems on par. Its API price is much cheaper and it's top up, not a subscription.”

“I've been fooling around with DeepSeek 4 agentically. It's probably not as good as Anthropic offerings, but even those seem to be roiled in politics and strife and DeepSeek 4 is very good IMHO.”

“Given deepseek-v4 is also the cheapest model among those five, I would say—based on these data—it's the winner.”

“DeepSeek V4 series, which uses an even more advanced attention mechanism and can save a lot of memory capacity for KV cache”

“Deepseek 4 is impressively cheap for the intelligence, but not reliable enough to daily drive”

“feels like between DeepSeek v4, then MiniMax M3, and now GLM-5.2 adding it 1M is rapidly becoming "table stakes"”

“Based solely on quality and price, OpenAI, Anthropic, and other western models just can't compete with the new generation of Chinese open models.”

“DeepSeek v4 which is Sonnet+ levels in my tests because it has an actual 1 million token context window”

“since DeepSeek v4 supports Claude Code's API you can literally set a few environment variables and continue to use Claude Code as a harness”

“I think it's great, but coming from Claude Code it did feel like going back in time by ~6 months in model capabilities.”

“spend a fraction in API billing (actually, significantly less than $100) while barely noticing a difference”

“open weigh models are good enough for many agentic tasks starting with Qwen 3.5/6 family and Deepseek v4 family”

“The models are open source open weights, which mean you can inspect what the models do and you you can choose US or EU infra providers”

“With the new DeepSeek V4 series and its uniquely memory-light KV cache you can even extend this to parallel inference”

“if you want frontier-level performance, as it was a few months back, Deepseek v4 and K2.6 are there”

“To me it feels like GLM 5.1, Kimi 2.6, DeepSeek 4 are all competitive both with each other and with the American models.”

“deep seek models are massively overrated in terms of how good they actually are at agantic usage”

“it did great, I would say on par with Sonnet 4.6 or even opus considering the thinking and explanation”

“OOS models are vool, cheap and promising for a future -- but why are we pretending they are better than they are?”

“DeepSeekV4 just came up with a new way to have super long context with KV cache an order of magnitude smaller than before.”

“It’s also not nearly as good as 5.5. I’ve had 3 different tasks just this week that deepseek has failed at”

“they'll get all that intellectual property beamed directly into their datacenters for analysis”

“data privacy, or how this AI may someday be used to undermine US national or economic security”

“without having to worry about token cost or about subscription limits or about any future degradation in performance”

“You might be underestimating how significantly cheaper this is and how much people care about price.”

“Especially if it increases sustainability and lowers the price, which this very clearly does.”

“open weights is really good, especially when base models (not just the instruction tuned) weights are released like here”

“These are open weights, not open source. They don't provide the training data or the methodology requires to reproduce these weights.”

“Open Source as it gets in this space, top notch developer documentation, and prices insanely low”

“they very obviously waited long enough for the GAB to add several dozen xz-level easter eggs to it”

“It is more than good enough and has effectively caught up with Opus 4.6 and GPT 5.4 according to the benchmarks.”

“This is free... as in you can download it, run it on your systems and finetune it to be the way you want it to be.”