DeepSeek V4 Flash

What hackers pick instead

• Claude OpusRecommended as a top coding model that fixes bugs and ships features with fewer mistakes than earlier LLMs, though critics cite weak architecture choices, usage limits and confident errors.3331recommend
• OpenAI GPTRecommended for strong coding, debugging and research help with good reasoning and token efficiency, though critics cite hallucinations, overcomplication, verbose output and high cost.1205recommend
• Claude SonnetA capable coding model that has improved noticeably with each version and is faster and cheaper than Opus, though it needs careful supervision and usage limits bite.854recommend
• FableRecommended as a top coding model for hard, long-running tasks and large refactors, though critics cite over-engineered code, costly usage limits and safety guards that switch models mid-task.823recommend
• Claude Fable 5Recommended for long-horizon agentic work and hard reasoning, debugging and research problems where it beats Opus, though overbroad safety filters often block legitimate security and biology tasks.774recommend

Every quote

The 40 most detailed of 1794 comments. Each links to the original.

1. it has proven so capable that after being simultaneously on Max x20 and Pro x20 subscriptions, i've transitioned to using DS 4.1 as a daily driver and am very satisfied

   Recommends · Sep 2026 · in “Livenerf: Has Opus 5.5 been nerfed yet?”

2. My personal favorite is DeepSeek v4.1 Flash but I have tried Qwen 3.8, Kimi 3 and GLM 5.3 which are equally impressive but DeepSeek is the cheapest and fastest

   Recommends · Sep 2026 · in “GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price”

3. I was running deepseek v4.1 pretty much non stop during work hours, with heavy tool/mcp usage and finding it very difficult to spend more than $75 in a month.

   Recommends · Oct 2026 · in “Why isn't the industry freaking out about DeepSeek 4.1 Flash?”

4. The flash version is great at implementing a plan but because it makes trade-offs for increased speed it doesn't take the time to think through everything

   Mixed · Oct 2026 · in “Why isn't the industry freaking out about DeepSeek 4.1 Flash?”

5. I have been loving deepseek flash. I can code in my hitl preference doing complex but moderate amount of coding and not even hit 1 dollar in a session.

   Recommends · Jul 2026 · in “DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis”

6. I dont even bother asking opus those kind of questions when I know deepseek-4.1-flash or glm-5.3* or mistral-large-4 will give the proper solution

   Recommends · Oct 2026 · in “I turned my unused Kindle into a photo frame with GPT-6 Astra”

7. I shifted from DeepSeek v4 Flash 0731 to Gemini 3.7 flash on openrouter and price shoots up almost double with no visible change in outcome.

   Recommends · Aug 2026 · in “GPT-5.6 Sol Pricing Cut by 50%”

8. I'm fine with DS Flash V4.1 in code bases I have experience in, don't really notice a huge difference other than more steering is needed.

   Recommends · Oct 2026 · in “Linus Torvalds's first LLM-assisted patch series”

9. In our tests, DeepSeek V4-flash is the cheapest model with acceptable replies (few hallucinations, while finding the right information).

   Recommends · May 2026 · in “DeepSeek makes the V4 Pro price discount permanent”

10. one-shotted by deepseek-v4.1-flsh-1017 (in high mode, not even max) with sound and programmatically generated assets in under 5 minutes

    Recommends · Oct 2026 · in “DeepSeek-v4.1-Flash-1017”

11. the capabilities are so far ahead anything else that the majority of the time it works really well and solves really complex problems

    Recommends · Jul 2026 · in “Bonsai 27B (1-bit LLM): The First 27B-Class Model to Run on a Phone”

12. I swear to god that DeepSeek V4-Flash is the most useful model available right now. It's SO FAST and is good enough for so many tasks

    Recommends · May 2026 · in “The last six months in LLMs in five minutes”

13. We can run several coding agents during the day and batch inference jobs overnight and serve the entire team with guaranteed privacy

    Recommends · Aug 2026 · in “DeepSeek V4 Flash on a Single AMD MI300X”

14. play with models that are cheap but bigger than what you'd reasonably be able to run locally like deepseek v4 flash (unquantized)

    Recommends · Jun 2026 · in “Qwen 3.6 27B is the sweet spot for local development”

15. DS Flash is where local models begin to feel useful for coding, but the quants we run locally are sharply reduced in intelligence

    Mixed · Aug 2026 · in “Qwen3.8-Max: A New Bar for Coding and Cowork”

16. tiny, highly cached, fast models like Qwen's 27b/a3b range or DeepSeek flash are highly capable and relatively inexpensive to run

    Recommends · Aug 2026 · in “Coding expertise is going to collapse from AI reliance”

17. Deepseek V4 Flash massively increased the value of local AI to me, to the point where it's displaced most of my Claude Code usage

    Recommends · Aug 2026 · in “Apple introduces M6 and M5 Ultra”

18. It's another eye-opener. If you run it in Claude Code, it's easy to forget you're not actually talking to a high-end Opus model.

    Recommends · Sep 2026 · in “MiMo v2.6”

19. I am running it as my assistant through Matrix with DeepSeek v4 Flash (and Qwen). It probably costs me about 2 dollars a month

    Recommends · Jul 2026 · in “Using an open model feels surprisingly good”

20. It's nice to be able to put deepseek flash on stupid, unnecessary or highly speculative tasks without thinking about the cost.

    Recommends · May 2026 · in “DeepSeek V4–almost on the frontier, a fraction of the price”

21. Coding tasks are lousy. Beyond coding, difficult open-ended reasoning scenarios are the ones at which it performs best for us.

    Mixed · Oct 2026

22. I've been using DS V4 Flash through OpenCode and it's mind-blowing how I get near-SOTA AI model for the price of two coffees.

    Recommends · Aug 2026 · in “AirLLM 70B inference with single 4GB GPU”

23. DeepSeek is then great because of the low price and price transparency, it just can’t compete with subsidized monthly access.

    Mixed · Oct 2026 · in “Why isn't the industry freaking out about DeepSeek 4.1 Flash?”

24. Open flash model is competing against OpenAI's 'Sonnet' model at the price of GPT 3, I am really excited about this release

    Recommends · Jul 2026 · in “DeepSeek-V4-Flash Update”

25. I am getting Opus 4.7 level performance at fraction of cost with DeepSeek V4 Flash set to high. It is a reliable workhorse.

    Recommends · Jul 2026 · in “DeepSeek-V4-Flash Update”

26. Deepseek especially given how cheap it is. It's also really solid and comparable in real world use to a sonnet for my work.

    Recommends · Sep 2026 · in “DeepSeek v4.1 Flash”

27. Rust works perfectly fine, but when I use Rust, I usually pay closer attention to performance, so I have to guide it a bit

    Mixed · Jul 2026 · in “DeepSeek-V4-Flash Update”

28. this model likely opens up near-frontier model inference (in slow, unattended scenarios) even on low-end consumer hardware

    Recommends · Sep 2026 · in “DeepSeek v4.1 Flash”

29. DeepSeek is amazing, they are, from a cost/benefit literally an order of magnitude or more better than the 'SOTA' models

    Recommends · Jul 2026 · in “DeepSeek-V4-Flash Update”

30. The bad news is that the original v4 flash was 284B, which was large but still somewhat reasonable for running locally.

    Mixed · Sep 2026 · in “DeepSeek v4.1 Flash”

31. I've been using Q4_K_S for a while now and it's been rock solid, but having tried Q3_XS, that would loop like crazy.

    Mixed · Oct 2026 · in “DeepSeek-v4.1-Flash-1017”

32. new DeepSeek V4.1 Flash 1017 really gets the numbers they promised for the first one, but it's too big for me to run

    Mixed · Oct 2026 · in “We're releasing the weights of Mistral Large 4”

33. V4 Flash 7/31 suffers from a lot of pathological behavior in coding harnesses as well, e.g. infinite loops, hallucinations, premature termination, and invalid tool calls.

    Criticizes · Sep 2026 · in “DeepSeek launching v4.1 flash cheaper and more capable than v4 pro”

34. deepseek-v4-flash tends to ignore instructions to use these tools and default to churning through code and guessing the cause of errors

    Criticizes · Jun 2026 · in “DeepSeek V4 Pro beats GPT-5.5 Pro on precision”

35. Flash makes a lot more initial mistakes, and then has to re-check stuff, and produces much more output compared to Pro.

    Criticizes · Aug 2026 · in “DeepSeek V4 Pro 0813”

36. I've definitely noticed that a high level of deepseek flash responses were looping endlessly before the 0731 release.

    Criticizes · Sep 2026 · in “So you want to use OpenRouter?”

37. The DS 0731 was so atrocious it was getting beaten by small Nemotron in my coding tests

    Criticizes · Oct 2026 · in “We're releasing the weights of Mistral Large 4”

38. KV cache optimizations in combination with the CED and compressed attention features are the reason why v4.1 flash has so few hallucination problems

    Recommends · Oct 2026 · in “Why isn't the industry freaking out about DeepSeek 4.1 Flash?”

39. That may be an issue with the harness you are using, i've never, and never heard of, someone having this problem specifically with this model.

    Recommends · Sep 2026 · in “DeepSeek launching v4.1 flash cheaper and more capable than v4 pro”

40. What's most impressive here is that that DeepSeek keeps showing how much of a performance improvement can come from post-training alone

    Recommends · Aug 2026 · in “DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis”