Deepseek did it again...
Check out MindsHub: https://tinyurl.com/5cra6js2 Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V Links: https://www.deepseek.com/en/news/deepseek-v4-1-flash/ https://x.com/The_Alex/status/2096294580006354985
Read Video · 文字起こしと深掘り分析
このエピソードには全文文字起こし + AI インサイトがあります
無料アカウント · カード不要 · 登録で150クレジット獲得、このエピソードのアンロックに十分
- 📄 タイムスタンプ付き全文文字起こし
- ✨ AI 要約・キーワード・マインドマップ
- 💡 重要ポイントと名言
エピソードのタイムライン
DeepSeek V4.1 Flash launches as a cheap, fast open-weights model that benchmarks near frontier models like Opus 5 and GPT 5.6 Soul.
- DeepSeek V4.1 Flash is incredibly cheap and fast, and benchmarks show it matching Opus 5 and GPT 5.6 Soul, continuing the pattern where open-weights models reach previous-generation frontier quality about six months behind the absolute frontier.
- The model is a 552 billion parameter mixture of experts with only 8 billion active parameters for input and 16 billion for output, making it relatively small compared to GPT 5.6's trillion parameters and Astra's estimated 7-10 trillion.
- On benchmarks it scores 30 on Terminal Bench 3.0 (only Opus 5 beats it) and 74.2 on Deep Suite, beating both Opus and GPT 5.6, though the host notes his own tests did not match the benchmark hype.
DeepSeek's efficiency gains slash memory requirements, cutting KV cache HBM needs to a quarter and SSD storage to an eighth.
- The KV cache only needs a fourth of the high bandwidth memory and an eighth of the SSD storage, which matters because HBM prices have been spiking from around $25 per gigabyte up to $10 per gigabyte as AI consumes all available memory.
- The memory footprint shrank eight times from DeepSeek V1 to V3.2, then 13 times smaller to V4 Flash, and another 4x smaller from V4 to V4.1, reflecting China's strength in making existing technology more efficient and faster rather than necessarily better.
- This efficiency is DeepSeek's answer to the global memory crunch, letting them serve many more tokens without paying for all that memory, and it shows up as extreme output speed reminiscent of the early Groq days.
Pricing is extremely low with peak and off-peak tiers, and the model is fully open weights so users can self-host.
- Off-peak pricing is 15 cents per million input tokens and 60 cents per million output tokens, doubling during peak hours to 30 cents input and $1.20 output, with cache hits costing a fraction of a penny.
- Because it is open source and open weights, users can download it, avoid giving DeepSeek their data, use any NeoCloud, or potentially run it locally once quantized if they have enough VRAM.
- The host contrasts this with frontier models like OpenAI and Anthropic charging around $50 per million output tokens, an orders-of-magnitude difference, though the vast majority of the economy only needs cheap models capable enough for 95% of use cases.
主要な概念
- DeepSeek V4.1 Flash— The newly released open-weights model that is the main subject of the episode.
- Mixture of Experts— Architecture that activates only a small subset of weights per query, enabling extreme efficiency.
- Efficiency gains— The core theme: DeepSeek's ability to cut memory and compute requirements dramatically.
注目の名言
And there's one big butt here. I've put it through a few tests and it didn't perform as well as I would have hoped, especially compared to the benchmarks, but I'm going to show you that later.
💡— Reveals a gap between benchmark scores and real-world performance, challenging the assumption that high benchmarks mean practical capability.
So of a 552 billion parameter model, only a tiny fraction of the weights are actually being used.
🤯— Highlights the extreme sparsity of mixture-of-experts models, showing how a huge model can run efficiently.
実行可能なテイクアウェイ
🤖AI Model Evaluation
Benchmarks can be misleading; real-world tests reveal practical weaknesses.
This week, pick a model you're considering and run a small custom test (e.g., a simple coding task) to verify its real capabilities.
Open-source models are catching up to frontier models within about six months.
Subscribe to a newsletter or set a reminder to check for new open-weight model releases every quarter.
💰Cost Optimization
Open models can be orders of magnitude cheaper than frontier models for most tasks.
Audit your current AI API usage and identify tasks that could be switched to a cheaper open model, then test one this week.
Off-peak pricing can significantly reduce costs for batch jobs.
Schedule non-urgent AI workloads to run during off-peak hours if your provider offers discounted rates.
文字起こしとAIインサイトは自動生成されたものであり、誤りが含まれる場合があります。認識精度は音質や話者の発話の明瞭さに左右されます——内容に不就がある場合は、元の音声をご確認ください。
すぐに読めるエピソードと動画
ポッドキャスト
INTJ Personality Type Advice - 0088
Personality Hacker Podcast
2015年10月19日1:05:25EN
A journalist's trick for talking to people you can't stand | Joshua Johnson
TED Talks Daily
2026年8月3日9:31EN
Humanity's First Star Probe, Architect Labs Beats NVIDIA 3.4x, Musk Wants Satellites to Cool Earth | EP #285
Moonshots with Peter Diamandis
2026年9月2日1:57:22EN
#8 「歴史と仲間と実力、全部欲しい」
朝井リョウ・加藤千恵 信頼できない語り手
2026年3月27日57:03JA
Vol.257 为什么上证是“综指”、深证用“成指”?
商业就是这样
2026年5月20日29:12ZH
Sind Deutschland Kinder egal? Mit Caroline von St. Ange
Politik mit Anne Will
2026年5月22日2:01:19DE
動画

Give Me 12 Minutes and I’ll Give You 30 Years of Productivity Advice
Daniel Pink
2025年9月14日11:59EN
GPUs, TPUs, & The Economics of AI Explained | Gavin Baker Interview
Invest Like The Best
2025年12月9日1:28:22EN
Neuroscientist: You Will NEVER Feel Stressed Again | Andrew Huberman
RESPIRE
2023年2月27日11:05EN
COMO SER GENTIL ESTÁ ACABANDO COM A SUA AUTORIDADE? | Fabiana Bertotti #152
Como Você Fez Isso?
2026年7月30日1:19:47PT
INDIA’S GOT LATENT S2 EP4 ft. Karan Aujla, Tanmay Bhat, Gurleen Pannu, Rahul Dua
Samay Raina
2026年8月2日53:54HI