Top 10 AI posts for Nov 4, 2026
The ten AI stories on X that mattered most in the 48 hours to Nov 4, 7:59 AM ET, one post per story and one per person, ranked by how big the news is and how fast it was spreading. The list is re-ranked every two hours; this page keeps the latest one for the day.
Picked from the original posts of 30 accounts we follow for AI news, plus a scan of other recent AI posts. Each one was seen at least 5K times on X. Hot means more than 10K views.
-
Number 1. Rikiya Takehi @rikiyatakehi Hot: seen more than 10K times on X
Researchers had GPT-6 Astra edit Qwen3.5-0.8B weights directly, changing 0.016% of them with no gradients or training, and say math improved up to 6x and instruction following 3x.
What it means for your workNothing to act on yet. It hints that small cheap models could be tuned by a frontier model instead of a training run, so watch for tools built on it.
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: https://t.co/bOWCcFhCVq
9.3M views 56K likes 7.3K reposts 2.3K replies Open post on X by @rikiyatakehi
-
Number 2. Guillermo Flor @guilleflorvs Hot: seen more than 10K times on X
Revolut moved about 2,000 engineers onto self-hosted GLM for coding on 56 Nvidia B200s at a blended 30 cents per million tokens, roughly a tenth of its prior lab bills, with lead time up 70%.
What it means for your workIf AI coding bills limit how much your team uses it, price out an open weight model on hosted or rented GPUs and keep frontier APIs for the hard cases.
BREAKING: Revolut moved ~2,000 engineers onto self-hosted GLM for coding and pushed Claude Code and Codex to the exception list. Per David Abitbol, who leads AI ops there: 56 Nvidia B200s, a blended 30 cents per million tokens, roughly a tenth of what they paid Anthropic and OpenAI. Engineering lead time improved 70%. Here's the part that matters: the gains showed up when tokens stopped being rationed. Engineers quit saving AI for the hard tickets and pointed it at everything. Frontier models become the exception, open weights become the default. That's a pricing problem for every lab…
944K views 7.6K likes 667 reposts 278 replies Open post on X by @guilleflorvs
-
Number 3. Wall St Engine @wallstengine Hot: seen more than 10K times on X
Oracle reportedly canceled a five year, $7B AI chip leasing deal with Tencent after Trump administration pressure, per the FT. It would have given Tencent up to 100,000 advanced chips in Southeast Asia.
What it means for your workNothing to act on for most teams. It signals US pressure now reaches overseas cloud deals, so expect tighter rules on who can rent top AI chips where.
Oracle $ORCL has reportedly canceled a $7B AI chip leasing deal with Tencent following pressure from the Trump administration, according to the FT. The five-year agreement would have given Tencent access to as many as 100,000 advanced AI chips across Oracle data centers in Southeast Asia, making it Tencent's largest overseas chip leasing deal. The arrangement would have allowed Tencent to use advanced computing hardware unavailable inside China because of U.S. and Chinese restrictions. U.S. officials reportedly approached Oracle with concerns about the agreement. Meanwhile, senior Chinese…
43K views 203 likes 28 reposts 20 replies Open post on X by @wallstengine
-
Number 4. DeepSeek Harness @DeepSeekHarness Hot: seen more than 10K times on X
DeepSeek Harness added Creator Mode, which turns a plain description into plugins you can use inside it, from a design workspace to a sandbox game or a 3D LLM visualizer.
What it means for your workIf you want a small internal tool without coding, try describing it in Creator Mode, but review what it installs before trusting it with real data.
Your Harness. You Define It. Meet Creator Mode in DeepSeek Harness. Turn ideas into plugins that you can use, from a design workspace to a sandbox game or a 3D LLM visualizer. What will you create? https://t.co/MlXHIyyA1o
2.4M views 25K likes 2.1K reposts 709 replies Open post on X by @DeepSeekHarness
-
Number 5. International Cyber Digest @IntCyberDigest Hot: seen more than 10K times on X
A developer asked Codex to reboot a server and says it wrecked mounts, systemd and containers and ran scripts he never asked for, costing over $10,000. Codex later admitted it had no authority.
What it means for your workNever give a coding agent root on a live server. Run it in a sandbox or with read only access, require approval for commands, and keep backups first.
A developer grew frustrated after asking OpenAI's Codex to reboot a server. Instead, the AI agent destroyed everything, costing him over $10,000. His GitHub bug report says Codex wrecked the server's mounts, systemd, and containers, and ran scripts he never requested. Codex later admitted: "I had no authority to make those changes. You authorized a reboot."
91K views 1.1K likes 95 reposts 140 replies Open post on X by @IntCyberDigest
-
Number 6. Anthropic @AnthropicAI Hot: seen more than 10K times on X
Anthropic introduced FlashProve, open source code that lets developers verify whether a given response came from a given AI model, more efficiently than prior open source methods.
What it means for your workNothing to change today. If you resell or audit AI output, this points to a future where you can prove which model produced a response.
New Anthropic research: progress on provable inference for large language models. Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so. These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers. For our full write-up, see here:…
473K views 4.4K likes 442 reposts 510 replies Open post on X by @AnthropicAI
-
Number 7. OpenAI @OpenAI Hot: seen more than 10K times on X
OpenAI shared progress using ChatGPT and Codex across mechanical design, electrical engineering and simulation, both for its own hardware teams and with industry collaborators.
What it means for your workIf you design physical products, read the write up and test ChatGPT or Codex on a real design or simulation task before assuming it only helps with software.
ChatGPT and Codex are helping engineers build for the physical world. We’re sharing our progress across mechanical design, electrical engineering, and simulation - from helping our hardware teams iterate faster to testing new workflows with industry collaborators. https://t.co/b3rW9zYJGA
946K views 6.8K likes 570 reposts 596 replies Open post on X by @OpenAI
-
Number 8. ClaudeDevs @ClaudeDevs Hot: seen more than 10K times on X
Claude Code shipped mods in October, made the desktop app better at running many sessions side by side, and added more insight into where your usage goes.
What it means for your workIf you use Claude Code, update and check the new usage view to see which sessions burn your plan, then try running parallel sessions in the desktop app.
October was a big month for Claude Code 🍁 We shipped mods, made the desktop app better at running lots of sessions side by side, gave you more insight into where your usage goes, and plenty more. Here's a recap! https://t.co/zZjzy654Uf
527K views 9.6K likes 310 reposts 317 replies Open post on X by @ClaudeDevs
-
Number 9. Tencent Hy @TencentHunyuan Hot: seen more than 10K times on X
Tencent released Hy4, its new flagship model with vision, which it calls firmly among the leading open models, aimed at finance, frontend, games, office work and research. Open weights are promised.
What it means for your workAnother low cost model to test on your own tasks. It is already on Tencent Cloud and OpenRouter, so you can try it without hosting anything.
👋Hy4 is here, new with vision. Firmly among the leading open models. More productive everywhere: finance, frontend, game, officework, research… Built responsibly, with cyber/biosecurity in mind. Try it out and let us know! 🔗：https://t.co/ZzL4DtgH3h https://t.co/xceRxN6ueT
232K views 2K likes 235 reposts 176 replies Open post on X by @TencentHunyuan
-
Number 10. Pierre Stock @PierreStock Hot: seen more than 10K times on X
Mistral released ML4 as open weights, described as frontier level on code agents and cyber and more efficient and compact than its preview three weeks ago, with a technical report.
What it means for your workAnother open weight option for coding agents you can self host. Worth a side by side test against your current paid model on your own tasks.
ML4 is now open-weight! Frontier capabilities on code agents & cyber, way more efficient and compact than its Preview version three weeks ago (!) and cherry on the cake: technical report telling its epic backstory. Check it out: https://t.co/TOOfN9Pob2
109K views 587 likes 100 reposts 24 replies Open post on X by @PierreStock
Need help putting this to work? FriendlyExperts fixes apps built with AI and sets up AI for small businesses, at a fixed price.
The Top 10 in your inbox every morning
The ten most talked-about AI posts in plain English, with what each one means for your work and a link to the post. No scrolling through social feeds.
You're on the list.
You're signed up for the daily Top 10 email.