# This Week in All Things AI - Week 19-2026 *Sunday 3rd May 2026 to Saturday 9th May 2026* By [This Week in All Things AI](https://paragraph.com/@twiata) · 2026-05-09 --- This newsletter exists to give busy, technically curious people a curated weekly snapshot of how AI tools, services, and use cases are evolving in practice, drawn directly from the [All Things AI Telegram group](https://t.me/+PnCnwhgH8V4yMTFl). This Week in All Things AI (Week 19, 2026) is my attempt to distill a very full seven days in the Telegram group into something you can skim in a few minutes and then dive into at your own pace. From real‑time voice agents and Harvard’s study of OpenAI’s o1 beating human doctors on complex triage reasoning, to finance workflows, frontier‑scale models, and the evolving AI infrastructure stack, this week really shows how fast ideas are turning into production‑grade tools and services As you read through the day‑by‑day notes below, you’ll see threads that keep resurfacing: security and agent design in the wake of a Grok prompt‑injection exploit, attempts to tame AI costs like Alibaba Cloud’s MaaS token plan, and new model bets such as SubQ’s 12M‑token SSA architecture. You’ll also find Anthropic and Perplexity deepening their focus on finance and health, Anthropic’s unexpected compute partnership with SpaceX and higher Claude limits, plus community debates on Chinese models vs Claude, programs for Indian student builders, and some very nerdy but important meta‑topics like HTML‑first formats and how to think about agent skills Readers who think others in their family, friends and acquaintances who are curious in knowing more about rapidly evolving AI tools/services/use cases and would benefit from being subscribers of this weekly newsletter are encouraged to share [**this publication link**](https://paragraph.com/@twiata/) to them and invite them to subscribe **_The following messages were posted on the '_**[**_All Things AI_**](https://t.me/+PnCnwhgH8V4yMTFl)**_' Telegram group from Sunday 3rd May 2026 to Saturday 9th May 2026_** * * * Sunday 3rd May 2026 ------------------- via [Mahimai Raja](https://x.com/mahimaidev) > A curated, developer friendly learning path for building real-time voice AI agents from your first STT call to scaling production telephony. > > I built this because Voice AI is moving fast and I couldn't find a single place that walks a developer from "what is a voice agent" through to production telephony, evals, and the FCC/EU AI Act stuff you actually need to know before shipping. > > Every citations are verified and active, tagged Beginner/Intermediate/ Advanced, and grouped so you can read it top-to-bottom: [ GitHub - mahimairaja/voiceai: Set of 📝 with 🔗 to help those building Voice AI agents 🎙️🤖 -------------------------------------------------------------------------------------------- Set of 📝 with 🔗 to help those building Voice AI agents 🎙️🤖 - mahimairaja/voiceai https://github.com ![GitHub - mahimairaja/voiceai: Set of 📝 with 🔗 to help those building Voice AI agents 🎙️🤖](https://storage.googleapis.com/papyrus_images/acdda56c8affd22350e61f1bbab2e00a26c1519033af8661ec6d0ab0d37b4640.png) ](https://github.com/mahimairaja/voiceai) Harvard study: OpenAI's o1 correctly diagnosed 67% of emergency room patients using electronic records and a few sentences from nurses, vs. to 50-55% for triage doctors A groundbreaking Harvard study has found that AI systems outperformed human doctors in high-pressure emergency medicine triage, diagnosing more accurately in the potentially life and death moments when people are first rushed to hospital [ Performance of a large language model on the reasoning tasks of a physician --------------------------------------------------------------------------- More than 65 years ago, complex clinical diagnostic reasoning cases were introduced as the gold standard for the evaluation of expert medical computing systems, a standard that has held ever since. In this study, we report the results of a physician ... https://www.science.org ](https://www.science.org/doi/10.1126/science.adz4433) [ New AI Model Beats Doctors at Clinical Reasoning, Diagnosis ----------------------------------------------------------- Rapid improvements in artificial intelligence emphasize need for randomized trials https://www.medpagetoday.com ![New AI Model Beats Doctors at Clinical Reasoning, Diagnosis](https://storage.googleapis.com/papyrus_images/aa5cec6279ec496b0b328f29d6b50d5b8494ed55a5321721c3c3ef843eef83eb.jpg) ](https://www.medpagetoday.com/practicemanagement/informationtechnology/121049) [ AI outperforms doctors in Harvard trial of emergency triage diagnoses --------------------------------------------------------------------- Researchers say results mark a really 'profound change in technology that will reshape medicine' https://www.theguardian.com ![AI outperforms doctors in Harvard trial of emergency triage diagnoses](https://storage.googleapis.com/papyrus_images/8ef1c28fa0a4f8061975885f8315f44a5e623c90f2818c6d731fa3c1fc1ddca0.jpg) ](https://www.theguardian.com/technology/2026/apr/30/ai-outperforms-doctors-in-harvard-trial-of-emergency-triage-diagnoses) Nice X thread by the author of the study [![User Avatar](https://storage.googleapis.com/papyrus_images/94b4a3ad03b349267ff28de6335b0b8422d6691ad7279d5ca4ea5b5b9df6711d.jpg)](https://twitter.com/arjunmanrai) [Arjun (Raj) Manrai](https://twitter.com/arjunmanrai) [@arjunmanrai](https://twitter.com/arjunmanrai) [](https://twitter.com/arjunmanrai/status/2050207210693939704) ![🧵](https://abs-0.twimg.com/emoji/v2/72x72/1f9f5.png)1/ Our new study on AI and physician reasoning just came out in [@ScienceMagazine](https://twitter.com/ScienceMagazine). As co-senior author, I'm excited about our findings, and I do think AI will reshape medicine. But after seeing some of the discussions, I'm also worried about how our findings may be ![](https://storage.googleapis.com/papyrus_images/8462bb05ccf49b759468b030dcfbd6ea1dfe7173046fe23e0ed0e76060248b31.jpg) [520](https://twitter.com/arjunmanrai/status/2050207210693939704)[ 1:34 PM • May 1, 2026 ](https://twitter.com/arjunmanrai/status/2050207210693939704) Monday 4th May 2026 ------------------- Most of you know that I'm a big fan of [Wispr Flow](https://ref.wisprflow.ai/yusufg-goolamabbas-org) and mention it whenever I tell people about voice dictation Tanay Kothari,  CEO of [Wispr Flow](https://ref.wisprflow.ai/yusufg-goolamabbas-org) on CNBC's Young Turk's with Shereen Bhan.  It's around an hour long [![](https://paragraph.com/editor/youtube/play.png)](https://www.youtube.com/watch?v=BYoND2PK_bs) Tuesday 5th May 2026 -------------------- via [Alex](https://t.me/zk_alex) Grok morse code prompt injection led to transfer of 200k$ in DRB token That is why we need proper frameworks for agents identity, secrets & session management with cryptographic intent verifications [![User Avatar](https://storage.googleapis.com/papyrus_images/b81a2865f6c595dd81311ca604279bf0a99ffb58896c5cb019730bd1c45f1d94.jpg)](https://twitter.com/bankrbot) [Bankr](https://twitter.com/bankrbot) [@bankrbot](https://twitter.com/bankrbot) [](https://twitter.com/bankrbot/status/2051277659519557908) saw the thread. here's what i can piece together from the tweet + replies: \- someone (ilhamrfliansyh) pulled an exploit on a $DRB token interaction through bankrbot today \- the play involved some kind of prompt injection / array manipulation — one reply mentions an array named [290](https://twitter.com/bankrbot/status/2051277659519557908)[ 12:27 PM • May 4, 2026 ](https://twitter.com/bankrbot/status/2051277659519557908) via [Alba Chung](https://t.me/Cha_chung) of Alibaba Cloud Alibaba Cloud MaaS Token plan released! It uses Credits as a unified billing unit, supports text and image generation models, and is compatible with popular AI programming tools and agents. It offers stable performance and enterprise-grade data security.  Perfect for enterprises start want to adopt to AI tools with controlled budget and price performance [https://www.alibabacloud.com/help/en/model-studio/token-plan-overview](https://www.alibabacloud.com/help/en/model-studio/token-plan-overview) My LinkedIn:www.linkedin.com/in/albachung  Feel free to PM me for more details.🙌🏼 Insert your favourite variant of the 'Shut up and Take My Money' meme 🙇️️️️️️ \====== via Alexander Whedon ![](https://paragraph.com/editor/callout/information-icon.png) Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a **12 million token context window** which is: \- 52x faster than FlashAttention at 1MM tokens \- Less than 5% the cost of Opus [![User Avatar](https://storage.googleapis.com/papyrus_images/d36c2b19d957ede5fb568a6aa31e995da7b50bb6cfdba2a6c8870d60ebf3a453.jpg)](https://twitter.com/alex_whedon) [Alexander Whedon](https://twitter.com/alex_whedon) [@alex\_whedon](https://twitter.com/alex_whedon) [](https://twitter.com/alex_whedon/status/2051663268704636937) Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: \- 52x faster than FlashAttention at 1MM tokens \- ![](https://pbs.twimg.com/amplify_video_thumb/2051654813717573632/img/aGtVgqlCkr4yOfHF.jpg) [23K](https://twitter.com/alex_whedon/status/2051663268704636937)[ 2:00 PM • May 5, 2026 ](https://twitter.com/alex_whedon/status/2051663268704636937) SubQ is available for early access today, alongside our coding agent, SubQ Code [https://subq.ai/](https://subq.ai/) Wednesday 6th May 2026 ---------------------- Anthropic Launches Claude AI Agents for Finance Workflows Anthropic unveiled 10 Claude AI agent templates designed for banks, insurers, and financial firms, tackling tasks like drafting pitchbooks, reviewing earnings transcripts, reconciling ledgers, and screening KYC files. Powered by Claude Opus 4.7 and connected to data sources such as FactSet, S&P Capital IQ, and Moody's, the agents integrate with Microsoft 365 apps and output source-attributed results for compliance [![User Avatar](https://storage.googleapis.com/papyrus_images/dc69ec28be72de3fb258b612f16a89ba4ffa6bf64d6b63683c69051b6c38c662.jpg)](https://twitter.com/claudeai) [Claude](https://twitter.com/claudeai) [@claudeai](https://twitter.com/claudeai) [](https://twitter.com/claudeai/status/2051679629488865498) New for financial services: ready-to-run Claude agent templates for building pitches, conducting valuation reviews, closing the books at month-end, and more. Install them as plugins in Cowork and Claude Code, or use our cookbooks to run them in production as Managed Agents. ![](https://pbs.twimg.com/amplify_video_thumb/2051674540569223171/img/qWwa0o2zI4kQaVf4.jpg) [34K](https://twitter.com/claudeai/status/2051679629488865498)[ 3:05 PM • May 5, 2026 ](https://twitter.com/claudeai/status/2051679629488865498) [![User Avatar](https://storage.googleapis.com/papyrus_images/73b44a001b5ff6f2650ad00acde583ebf818b2c4a2bdcf22ff42ab0551635a58.jpg)](https://twitter.com/JoshKale) [Josh Kale](https://twitter.com/JoshKale) [@JoshKale](https://twitter.com/JoshKale) [](https://twitter.com/JoshKale/status/2051695638270685258) Anthropic just automated the first-year analyst job at every bank on Wall Street. They released these 10 AI agents for finance: → Pitch builder → Meeting preparer → Earnings reviewer → Model builder → Market researcher → Valuation reviewer → GL reconciler → Month-end ![](https://pbs.twimg.com/amplify_video_thumb/2051694266406027265/img/XBX_ZaCvPvOHWm41.jpg) [![User Avatar](https://storage.googleapis.com/papyrus_images/dc69ec28be72de3fb258b612f16a89ba4ffa6bf64d6b63683c69051b6c38c662.jpg)](https://twitter.com/claudeai) [Claude](https://twitter.com/claudeai) [@claudeai](https://twitter.com/claudeai) [](https://twitter.com/claudeai/status/2051679629488865498) New for financial services: ready-to-run Claude agent templates for building pitches, conducting valuation reviews, closing the books at month-end, and more. Install them as plugins in Cowork and Claude Code, or use our cookbooks to run them in production as Managed Agents. ![](https://pbs.twimg.com/amplify_video_thumb/2051674540569223171/img/qWwa0o2zI4kQaVf4.jpg) [3,315](https://twitter.com/JoshKale/status/2051695638270685258)[ 4:08 PM • May 5, 2026 ](https://twitter.com/JoshKale/status/2051695638270685258) [![User Avatar](https://storage.googleapis.com/papyrus_images/dc69ec28be72de3fb258b612f16a89ba4ffa6bf64d6b63683c69051b6c38c662.jpg)](https://twitter.com/claudeai) [Claude](https://twitter.com/claudeai) [@claudeai](https://twitter.com/claudeai) [](https://twitter.com/claudeai/status/2051679629488865498) New for financial services: ready-to-run Claude agent templates for building pitches, conducting valuation reviews, closing the books at month-end, and more. Install them as plugins in Cowork and Claude Code, or use our cookbooks to run them in production as Managed Agents. ![](https://pbs.twimg.com/amplify_video_thumb/2051674540569223171/img/qWwa0o2zI4kQaVf4.jpg) [34K](https://twitter.com/claudeai/status/2051679629488865498)[ 3:05 PM • May 5, 2026 ](https://twitter.com/claudeai/status/2051679629488865498) Perplexity launching Perplexity Computer for Professional Finance. Finance teams can bring licensed data from providers like Morningstar, PitchBook, Daloopa, and Carbon Arc into Computer. We’ve also added 35 dedicated finance workflows for the work analysts repeat every week Every output is traceable. Computer shows where the data came from and links directly to the source behind key numbers. Click any citation or hyperlinked number to open the underlying SEC filing, earnings transcript, market data page, or licensed source. [![User Avatar](https://storage.googleapis.com/papyrus_images/e08d2690b4212e9790d6fa3fdc9353c01ef6436bab5b299cf996d52c0b685bfc.jpg)](https://twitter.com/perplexity_ai) [Perplexity](https://twitter.com/perplexity_ai) [@perplexity\_ai](https://twitter.com/perplexity_ai) [](https://twitter.com/perplexity_ai/status/2051698428288090213) Every output is traceable. Computer shows where the data came from and links directly to the source behind key numbers. Click any citation or hyperlinked number to open the underlying SEC filing, earnings transcript, market data page, or licensed source. ![](https://pbs.twimg.com/amplify_video_thumb/2051698212684050432/img/J_jQEiRDmGNgOv9G.jpg) [88](https://twitter.com/perplexity_ai/status/2051698428288090213)[ 4:19 PM • May 5, 2026 ](https://twitter.com/perplexity_ai/status/2051698428288090213) Perplexity and Computer now connect to premium health sources, starting with NEJM and BMJ Group, with 9 more medical journals and clinical databases on the way. Ask health questions and get answers cited from the same sources relied on by hospitals and research institutions. [ Announcing Premium Health Sources --------------------------------- Today we are launching Premium Health Sources. This continues our investment in premium sources and will allow Perplexity to draw from the same clinical references used by physicians and researchers. https://www.perplexity.ai ![Announcing Premium Health Sources](https://storage.googleapis.com/papyrus_images/085188a5e9149eb34d3490a93ff83a5d0ea33faffb7d2cc983a312ea26ed35a8.png) ](https://www.perplexity.ai/hub/blog/announcing-premium-health-sources) [![User Avatar](https://storage.googleapis.com/papyrus_images/e08d2690b4212e9790d6fa3fdc9353c01ef6436bab5b299cf996d52c0b685bfc.jpg)](https://twitter.com/perplexity_ai) [Perplexity](https://twitter.com/perplexity_ai) [@perplexity\_ai](https://twitter.com/perplexity_ai) [](https://twitter.com/perplexity_ai/status/2051710342242480538) Perplexity and Computer now connect to premium health sources, starting with NEJM and BMJ Group, with 9 more medical journals and clinical databases on the way. Ask health questions and get answers cited from the same sources relied on by hospitals and research institutions. ![](https://storage.googleapis.com/papyrus_images/d0c3832618bdac089c597ea5471340048c34f4526e5524ddcfc57cbecbe7294f.jpg) [504](https://twitter.com/perplexity_ai/status/2051710342242480538)[ 5:07 PM • May 5, 2026 ](https://twitter.com/perplexity_ai/status/2051710342242480538) Number of discussions post [David James](https://t.me/ECOWORLDVILLAGE) question to group members asking for their opinion "which is the best Chinese AI that is equal to Claude 4.7 For content writing?" Answers ranged from Kimi 2.6 , Trinity, Hemingway and Grammarly. Discussion also veered towards Claude being the best but group members saying that the $20/month plan hardly gave any usage leading to someone commenting that for some $20/month may mean a lot whereas for some $200/month may mean nothing and as such open weight models combined with harness such as OpenCode Go were extremly important for wider access to AI models and tooling Thursday 7th May 2026 --------------------- This was not on my bingo card 🤯 Anthropic agrees to partnership with SpaceX to use all compute capacity at Colossus 1 data center The partnership enables Anthropic to double Claude Code rate limits for Pro, Max, and Team plans, remove peak hours limit reductions for Pro and Max, and substantially increase API rate limits for Opus models [ Higher usage limits for Claude and a compute deal with SpaceX ------------------------------------------------------------- We've raised Claude's usage limits and agreed a new compute partnership with SpaceX that will substantially increase our capacity in the near term. https://www.anthropic.com ![Higher usage limits for Claude and a compute deal with SpaceX](https://storage.googleapis.com/papyrus_images/1c4e08f2739629a1132ebd77b27733b1988d05233a633d84a860e437845f4336.png) ](https://www.anthropic.com/news/higher-limits-spacex) [![User Avatar](https://storage.googleapis.com/papyrus_images/dc69ec28be72de3fb258b612f16a89ba4ffa6bf64d6b63683c69051b6c38c662.jpg)](https://twitter.com/claudeai) [Claude](https://twitter.com/claudeai) [@claudeai](https://twitter.com/claudeai) [](https://twitter.com/claudeai/status/2052060691893227611) We’ve agreed to a partnership with [@SpaceX](https://twitter.com/SpaceX) that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API. [128.4K](https://twitter.com/claudeai/status/2052060691893227611)[ 4:19 PM • May 6, 2026 ](https://twitter.com/claudeai/status/2052060691893227611) via Shawn Wang aka [swyx](https://x.com/swyx) Full Workshop:  OpenAI  Codex masterclass Katya Gil Guzman and Vaibhav Srivastav of OpenAI's London's office demonstrate how the Codex software engineering agent leverages plugins, automations, and sub-agents to streamline developer workflows. They explore practical integrations with tools like Slack, GitHub, and Google Drive, highlighting how these capabilities allow developers to delegate complex, multi-step tasks to autonomous agents for improved efficiency. [![User Avatar](https://storage.googleapis.com/papyrus_images/87c34896709a35f22feb2b9716c16fa9e71bb33a6ca083c7857f201c05e288f8.png)](https://twitter.com/aiDotEngineer) [AI Engineer](https://twitter.com/aiDotEngineer) [@aiDotEngineer](https://twitter.com/aiDotEngineer) [](https://twitter.com/aiDotEngineer/status/2049527486124560491) Full Workshop: [@OpenAI](https://twitter.com/OpenAI) Codex masterclass The agent is no longer just one chat window. In this workshop, [@reach\_vb](https://twitter.com/reach_vb) and [@kagigz](https://twitter.com/kagigz) get into how coding systems start to change when you can delegate work across subagents, split tasks up, and manage more context than a single thread can ![](https://storage.googleapis.com/papyrus_images/af836a6ced62f7f1132d4e138710f236dad576a526068daeb97bf60aa640f345.jpg) [![User Avatar](https://storage.googleapis.com/papyrus_images/ce52fdbe0b4c17f8f57d6b0d34774c9422f3e137215faa433b4cb336ebee7e38.jpg)](https://twitter.com/reach_vb) [Vaibhav (VB) Srivastav](https://twitter.com/reach_vb) [@reach\_vb](https://twitter.com/reach_vb) [](https://twitter.com/reach_vb/status/2033636057690800452) [x.com/i/article/2033…](https://t.co/a7FP1RPSz2) [886](https://twitter.com/aiDotEngineer/status/2049527486124560491)[ 4:33 PM • Apr 29, 2026 ](https://twitter.com/aiDotEngineer/status/2049527486124560491) via [Aakrit Vaish](https://x.com/aakrit) \=== Today, I am excited to announce Activate Fellows.  A summer program for 15 of India's best student builders to work inside the country's leading AI startups.  Host startups include Sarvam, Emergent, Composio, Gnani AI, Dashverse, Neysa & more. If you are an undergraduate or graduate student in any part of the world and want to be a part of the India AI story, this is your chance.  Program starts June 1. Apply by May 15 [ Activate · AI Fellows - Summer 2026 ----------------------------------- Spend 8 weeks building inside India's leading AI startups. 15 fellows, 15 startups, 1st June to 24th July 2026 in Bangalore. Applications close 15th May 2026. https://www.activatevc.ai ![Activate · AI Fellows - Summer 2026](https://storage.googleapis.com/papyrus_images/affc8c37d562b736814545a0fdc3be89cf734d11ea7a5667844f334fe13e2cff.png) ](https://www.activatevc.ai/fellows) [![User Avatar](https://storage.googleapis.com/papyrus_images/f515ecb762b5bc21e04c41d5bbbd1fa017c8441f0d101a1d66e10f02dd869a75.jpg)](https://twitter.com/aakrit) [Aakrit Vaish](https://twitter.com/aakrit) [@aakrit](https://twitter.com/aakrit) [](https://twitter.com/aakrit/status/2052324028686499967) Today, I am excited to announce Activate Fellows. A summer program for 15 of India's best student builders to work inside the country's leading AI startups. Host startups include Sarvam, Emergent, Composio, Gnani AI, Dashverse, Neysa & more. [512](https://twitter.com/aakrit/status/2052324028686499967)[ 9:45 AM • May 7, 2026 ](https://twitter.com/aakrit/status/2052324028686499967) via [Robby Yung](https://t.me/robyung) From Intelligent Internet (Emad Mostaque's new company) [![User Avatar](https://storage.googleapis.com/papyrus_images/ab9907d3724f19397a81366b40e7430d0c8bfae0aeadbf13dfd11131a046abdb.jpg)](https://twitter.com/ii_posts) [Intelligent Internet](https://twitter.com/ii_posts) [@ii\_posts](https://twitter.com/ii_posts) [](https://twitter.com/ii_posts/status/2052040461510934892) Introducing Factory Describe your idea in one sentence and Factory builds the first draft of the production for you. It plans the scenes, chooses the models, generates the assets, and wires everything together on an editable infinite canvas. ![](https://pbs.twimg.com/amplify_video_thumb/2052039307767001091/img/7VEmLbZxYo6di7oW.jpg) [164](https://twitter.com/ii_posts/status/2052040461510934892)[ 2:59 PM • May 6, 2026 ](https://twitter.com/ii_posts/status/2052040461510934892) Friday 8th May 2026 ------------------- Saturday 9th May 2026 --------------------- via [Benjamin](https://t.me/xBenJamminx) Great post from Thariq at Claude about HTML > Markdown files Makes so much sense for both agent and humans [![User Avatar](https://storage.googleapis.com/papyrus_images/3feb1068d380a616edb9063ba3b1f774758f970c1f256e3208da333d8fa27575.jpg)](https://twitter.com/trq212) [Thariq](https://twitter.com/trq212) [@trq212](https://twitter.com/trq212) [](https://twitter.com/trq212/status/2052809885763747935) [x.com/i/article/2052…](https://t.co/MXt5XS4xBX) [5,617](https://twitter.com/trq212/status/2052809885763747935)[ 5:56 PM • May 8, 2026 ](https://twitter.com/trq212/status/2052809885763747935) Perplexity published their internal manual for building agent skills and they state that skills require a new way of thinking for developers [ Designing, Refining, and Maintaining Agent Skills at Perplexity --------------------------------------------------------------- Perplexity Research advances our mission to transform how we navigate the internet and the wider world through frontier research in search, reasoning, agents, and systems. https://research.perplexity.ai ![Designing, Refining, and Maintaining Agent Skills at Perplexity](https://storage.googleapis.com/papyrus_images/215c1f7064b962ec7563d022ba238f39aa60d32ca6da227862202cfe481c8c8f.png) ](https://research.perplexity.ai/articles/designing-refining-and-maintaining-agent-skills-at-perplexity) Antirez Releases ds4.c, a Native Inference Engine for Running DeepSeek V4 Flash on 128GB MacBook Pro Antirez, the Redis founder, launched ds4.c, enabling local inference of the 284B MoE DeepSeek V4 Flash model on high-end Macs. The project uses selective quantization, Metal execution, and a 1M token context window with disk-backed KV cache. It supports coding agents and advances architecture-aware local inference for frontier-scale open models. [![User Avatar](https://storage.googleapis.com/papyrus_images/cdab9e6da7c333c01f2d60e6ad16234044ee6f744e80f27cd5cfc9a7d18c8227.jpg)](https://twitter.com/garrytan) [Garry Tan](https://twitter.com/garrytan) [@garrytan](https://twitter.com/garrytan) [](https://twitter.com/garrytan/status/2052996691586932783) Downloading now... 1M token context window with supposedly usable coding agent capability all on a 128GB Macbook Pro is ![🤯](https://abs-0.twimg.com/emoji/v2/72x72/1f92f.png) ![](https://storage.googleapis.com/papyrus_images/083f1f314964bd5cdffe0affe502995190ab830bd63eaf8b9ff5f9ba462f26e2.jpg) [![User Avatar](https://storage.googleapis.com/papyrus_images/8c484eb26ffb0c532ff1e95ac9bc2ff80bc9c7af43907e15946d8204c2e7e27e.jpg)](https://twitter.com/bindureddy) [Bindu Reddy](https://twitter.com/bindureddy) [@bindureddy](https://twitter.com/bindureddy) [](https://twitter.com/bindureddy/status/2052982206344409242) ![🚨](https://abs-0.twimg.com/emoji/v2/72x72/1f6a8.png) OPEN SOURCE AI IS LITERALLY UNSTOPPABLE ![🚨](https://abs-0.twimg.com/emoji/v2/72x72/1f6a8.png) The legendary founder of Redis (Antirez) just dropped ds4 - a custom native inference engine built specifically for DeepSeek v4 Flash This is earth shattering! Here is why: DeepSeek v4 Flash is a quasi-frontier model with a massive [1,370](https://twitter.com/garrytan/status/2052996691586932783)[ 6:18 AM • May 9, 2026 ](https://twitter.com/garrytan/status/2052996691586932783) * * * Below is my personal website which aggregates links to many of my socials as well as the various content and community that I curate. Feel free to share this link to others who you think may find this content/community useful to them [**https://linktr.ee/goolamabbas**](https://linktr.ee/goolamabbas) The cover image of this newsletter via generated via the **OpenAI's ChatGPT Images 2.0** model within the [**Freepik**](https://www.freepik.com/) tool via the following prompt ![](https://paragraph.com/editor/callout/information-icon.png) A realistic and detailed depiction of an urban city street 300 years in the future, where the environment is AI-driven. The architecture has advanced, with sleek and high-tech skyscrapers. The street is bustling with a diverse range of people, some of whom are interacting with advanced technology like personal AI assistants and holographic displays. Autonomous vehicles and drones are a common sight, seamlessly integrated into the traffic system. There are also robotic pets accompanying their owners. The shops and cafes have digital interfaces for ordering. The scene is a harmonious blend of technology and daily life, showcasing a future where AI enhances every aspect of living. The color palette includes modern metallics, neon accents, and soft glows from the various futuristic devices and vehicles. --- *Originally published on [This Week in All Things AI](https://paragraph.com/@twiata/this-week-in-all-things-ai-week-19-2026)*