# This Week in All Things AI - Week 34-2026

*Sunday 16th August 2026 to Saturday 22nd August 2026*

By [This Week in All Things AI](https://paragraph.com/@twiata) · 2026-08-23

---

This Week in All Things AI covers key developments in models, agents, tools, infrastructure, and policy curated from discussions in the [**All Things AI Telegram group.**](https://t.me/+PnCnwhgH8V4yMTFl)

If you follow AI for work, research, investing, or just to understand where the technology is heading, this [**weekly brief**](https://paragraph.com/@twiata) is a concise way to scan the most important launches, risks, and resources in a few focused minutes.

The week of 16th to 22nd August 2026 brought a strong mix of multimodal model releases, agent infrastructure, developer tools, and practical AI adoption. RedNote released dots3-note preview, an open-weight 280B MoE model built for long-horizon multimodal agents, while DeepSeek introduced V4-Flash-Vision-Exp for visual agent workloads and OpenRouter debuted the mysterious, high-performing Ox Alpha coding model. On the product side, Cartesia shipped Sonic-3.6 voice AI, Wispr raised a $280M Series B, and Meta expanded its AI offering with a native Mac app and business-oriented integrations.

The week also highlighted the rapidly maturing ecosystem around AI operators: Artificial Analysis launched a benchmark for agent web-search APIs; K-Dense released drug-discovery agent skills; Harvey opened a legal-engineering certification; and Replit introduced a credit-free mode for routine agent work. Google made AI plans free for eligible students across more than 140 countries, while Bun 1.4 delivered a major Rust-based runtime overhaul for JavaScript and TypeScript developers. Alongside the releases, Anthropic CEO Dario Amodei’s public defence of AI-risk discussions and his optimism on AI-enabled biomedical progress underscored the continuing tension between capability, adoption, and safety.

The sections that follow walk through these items day by day, with short context and links so you can dive deeper into the pieces most relevant to your work or interest

* * *

Sunday 16th August 2026
-----------------------

Xiaohongshu/RedNote announced dots3-note preview, an open-weight multimodal MoE AI model from RedNote's AI lab aimed at advancing long-horizon agency for real-life tasks like reasoning, environment exploration, memory updates, and tool use. 

Key specs include 280B total parameters with 16B active, 512K context window, native support for text/vision/video/audio, and TEMPO—a new RL method using self-critiquing and test-time scaling for extended agent training. 

Accompanying benchmark charts show competitive performance against larger models on agentic, coding, and multimodal tasks; model weights are on Hugging Face, with API access now available via OpenRouter. 

[![User Avatar](https://storage.googleapis.com/papyrus_images/87aef67f05526be78f32234199b0dd2e8d959b5b60f5427cc677022347bd7479.png)](https://twitter.com/dotsstudioai)

[dots studio](https://twitter.com/dotsstudioai)

[@dotsstudioai](https://twitter.com/dotsstudioai)

[](https://twitter.com/dotsstudioai/status/2088083314855018521)

Introducing dots3-note preview — a small but mighty step toward long-horizon agency in real life.  
![🔹](https://abs-0.twimg.com/emoji/v2/72x72/1f539.png) 280B MoE with 16B active parameters, a 512K context window, and multimodal understanding across text, vision, and audio  
![🔹](https://abs-0.twimg.com/emoji/v2/72x72/1f539.png) Introduces TEMPO, a new RL approach for long-horizon

![](https://storage.googleapis.com/papyrus_images/e6cd62ce30f23d0892897365bd7e6351acdd9b0d72ef1fb25e52eebe550fd1a5.jpg)

[1,562](https://twitter.com/dotsstudioai/status/2088083314855018521)[

2:00 AM • Aug 14, 2026

](https://twitter.com/dotsstudioai/status/2088083314855018521)

[![User Avatar](https://storage.googleapis.com/papyrus_images/d6d6c09c6a47b31a8252c4f3cffde2a96b3f1b84a6a623dc5d97405387417981.jpg)](https://twitter.com/ChaoQiao42)

[Chao Qiao](https://twitter.com/ChaoQiao42)

[@ChaoQiao42](https://twitter.com/ChaoQiao42)

[](https://twitter.com/ChaoQiao42/status/2088004290556436991)

WE SHIPPED IT! ![🚀](https://abs-0.twimg.com/emoji/v2/72x72/1f680.png)  
  
Introducing dots3-note Preview: our open-source 280B/16B-active multimodal model, built for complex reasoning and long-horizon agents.  
  
Beyond the model, we’re sharing TEMPO (Test-time-scaled Value Estimation with Macro-step Policy Optimization)—our answer to

![](https://storage.googleapis.com/papyrus_images/bbe6690dbf04da76ad7479aafb5fb3774d880ff3d2ca2da49130290d21baadea.jpg)

[1,581](https://twitter.com/ChaoQiao42/status/2088004290556436991)[

8:46 PM • Aug 13, 2026

](https://twitter.com/ChaoQiao42/status/2088004290556436991)

[![User Avatar](https://storage.googleapis.com/papyrus_images/d6d6c09c6a47b31a8252c4f3cffde2a96b3f1b84a6a623dc5d97405387417981.jpg)](https://twitter.com/ChaoQiao42)

[Chao Qiao](https://twitter.com/ChaoQiao42)

[@ChaoQiao42](https://twitter.com/ChaoQiao42)

[](https://twitter.com/ChaoQiao42/status/2088367822741406147)

Seeing dots3-note finally open-sourced is deeply personal for me.  
  
Over the past months, our post-training team poured an extraordinary amount of work into the model as a whole—reasoning, coding, tool use, multimodal understanding, and long-horizon agency. Behind the results were

[![User Avatar](https://storage.googleapis.com/papyrus_images/d6d6c09c6a47b31a8252c4f3cffde2a96b3f1b84a6a623dc5d97405387417981.jpg)](https://twitter.com/ChaoQiao42)

[Chao Qiao](https://twitter.com/ChaoQiao42)

[@ChaoQiao42](https://twitter.com/ChaoQiao42)

[](https://twitter.com/ChaoQiao42/status/2088366133279867044)

How do you train an agent when a single rollout takes tens of hours?  
  
Alongside the open-source release of dots3-note Preview, we’re sharing more details about TEMPO: Test-Time-Scaled Value Estimation with Macro-Step Policy Optimization.  
  
Long-horizon agent RL faces two

![](https://storage.googleapis.com/papyrus_images/3ed392d2d7c2a5d6d02d1872fba1209e25971ec871d437fbd7283e7d03c65d67.jpg)

[6](https://twitter.com/ChaoQiao42/status/2088367822741406147)[

8:51 PM • Aug 14, 2026

](https://twitter.com/ChaoQiao42/status/2088367822741406147)

Anthropic CEO Defends AI Risk Talk While Eyeing Disease Cures

In a detailed X exchange, Anthropic CEO Dario Amodei pushed back on criticism from investor Gavin Baker that his risk warnings fuel anti-AI views. Amodei stressed that real breakthroughs, like curing most human diseases in 5-10 years, will outweigh messaging, and revealed Anthropic's push into biology and medicine with early results expected soon. He defended regulation proposals that aid smaller players while tackling cyber, bio, and alignment risks, as prediction markets buzz with trader bets on his timeline

[![User Avatar](https://storage.googleapis.com/papyrus_images/71aeb71e3b14fd5a361508115871b52614a5d38c8489c4a8e562841a7b5e125d.jpg)](https://twitter.com/DarioAmodei)

[Dario Amodei](https://twitter.com/DarioAmodei)

[@DarioAmodei](https://twitter.com/DarioAmodei)

[](https://twitter.com/DarioAmodei/status/2088758816376807762)

1/2 Thanks Gavin for an especially thoughtful exchange. I don't usually spend much time on social media but I wanted to engage here because it really brings out the heart of an important conversation.  
  
First, on regulation, I think that “either concentrate it in the hands of a

[![User Avatar](https://storage.googleapis.com/papyrus_images/e3a46823fc61c026eead44cd80722218d38f2d837988a7b92e5dea2aa371367b.jpg)](https://twitter.com/GavinSBaker)

[Gavin Baker](https://twitter.com/GavinSBaker)

[@GavinSBaker](https://twitter.com/GavinSBaker)

[](https://twitter.com/GavinSBaker/status/2088611616577253502)

Sholto, thank you for setting the record straight. Larger issue is that multiple very serious people in Silicon Valley have heard some variation of this and believe it to be true. And the reason it is believable to so many is that it is consistent with Dario’s public messaging

[8,917](https://twitter.com/DarioAmodei/status/2088758816376807762)[

10:44 PM • Aug 15, 2026

](https://twitter.com/DarioAmodei/status/2088758816376807762)

[![User Avatar](https://storage.googleapis.com/papyrus_images/f1eb2c134dc601a062329bd2af7f45aefe6af1ec1b66c0da6a133c3d23082836.jpg)](https://twitter.com/_sholtodouglas)

[Sholto Douglas](https://twitter.com/_sholtodouglas)

[@\_sholtodouglas](https://twitter.com/_sholtodouglas)

[](https://twitter.com/_sholtodouglas/status/2088463770318516734)

Completely false.  
  
I like Gavin's takes, but whoever he heard this from is lying so that it fits the narrative some people so desperately want you to believe.  
  
The same people will try to convince you Anthropic has no moat, and a sentence later that it might become so powerful

![](https://storage.googleapis.com/papyrus_images/28d9306e4e6d04b921db208032f92c72827c71ed99b05305d91c4834a97b6c7a.jpg)

[![User Avatar](https://storage.googleapis.com/papyrus_images/569d3f77739b04c0530013b4ef6b5ea116773eaa72dec764790bb907a0e5e027.jpg)](https://twitter.com/theallinpod)

[The All-In Podcast](https://twitter.com/theallinpod)

[@theallinpod](https://twitter.com/theallinpod)

[](https://twitter.com/theallinpod/status/2088367978270142811)

Gavin Baker: Anthropic Believes They Could Be the ONLY Company Left in the World  
  
“I would certainly discourage Dario from saying that ever again to anyone.”  
  
[@GavinSBaker](https://twitter.com/GavinSBaker):  
  
“Internally, Anthropic is very confident. I have been told by multiple people I trust that Dario has said

![](https://pbs.twimg.com/amplify_video_thumb/2088366674667773952/img/sBriUBHyBAaQ-W9h.jpg)

[2,278](https://twitter.com/_sholtodouglas/status/2088463770318516734)[

3:12 AM • Aug 15, 2026

](https://twitter.com/_sholtodouglas/status/2088463770318516734)

Monday 17th August 2026
-----------------------

Voice AI startup [Wispr](https://ref.wisprflow.ai/yusufg-goolamabbas-org) raised a $280M Series B led by Menlo Ventures at a $2B valuation, taking its total funding to $361M, and now has 100K business customers

[

Wispr talks its way to a $2 billion valuation. The company says dictation's only the beginning | Fortune
--------------------------------------------------------------------------------------------------------

Wispr has raised $280 million in its Menlo Ventures-led Series B, as it looks beyond dictation.

https://fortune.com

![Wispr talks its way to a $2 billion valuation. The company says dictation's only the beginning | Fortune](https://storage.googleapis.com/papyrus_images/f24d98fc4fd029ef5df1b1119a5663c4b818d964f5675f37dc52447596af63e9.png)

](https://fortune.com/2026/08/17/wispr-2-billion-valuation-dictations-only-the-beginning/)

[![User Avatar](https://storage.googleapis.com/papyrus_images/7373f69ed9fe07e31232dc06d08d94a0c12a15c1d069c2841933497efb1cd749.jpg)](https://twitter.com/vibinbaburajan)

[Vibin Babuurajan 👋](https://twitter.com/vibinbaburajan)

[@vibinbaburajan](https://twitter.com/vibinbaburajan)

[](https://twitter.com/vibinbaburajan/status/2080178485507784769)

Wispr Flow got into talks to raise $260M after VCs had already been using the product for months.  
  
As far as the story goes, Wispr Flow didn't reach out to Menlo Ventures.  
  
A partner there had already been dictating his emails and memos through the app for months.

![](https://storage.googleapis.com/papyrus_images/8a75ce8d53eb4b10a633fab5e2b83f7a0602aa5332eae2a648cd692b563ff783.jpg)

[21](https://twitter.com/vibinbaburajan/status/2080178485507784769)[

6:29 AM • Jul 23, 2026

](https://twitter.com/vibinbaburajan/status/2080178485507784769)

Tuesday 18th August 2026
------------------------

Harvey, an AI company specializing in complex legal work, announced the launch of its free self-paced Legal Engineering Certification via Harvey Academy. 

The course teaches four core skills—legal expertise, AI fluency, process design, and user enablement—to help professionals design AI-enabled legal workflows and drive adoption. 

Aimed at lawyers, in-house teams, and legal tech roles, it includes practical training from over 1,000 AI implementations and awards a shareable badge upon completion. 

[![User Avatar](https://storage.googleapis.com/papyrus_images/09fb7c99aafb07dc95a57a530edd93ff56cdc354581adc448a6dfeab95f0f51e.jpg)](https://twitter.com/harvey)

[Harvey](https://twitter.com/harvey)

[@harvey](https://twitter.com/harvey)

[](https://twitter.com/harvey/status/2089321466320023825)

Now live: Harvey's Legal Engineering Certification.  
  
A free, self-paced course covering core legal engineering skills: legal expertise, AI fluency, process design, and user enablement.  
  
Open to everyone - no Harvey account required.

![](https://pbs.twimg.com/amplify_video_thumb/2089321410774872064/img/aZJEDIneSBJEx-I4.jpg)

[211](https://twitter.com/harvey/status/2089321466320023825)[

12:00 PM • Aug 17, 2026

](https://twitter.com/harvey/status/2089321466320023825)

Cartesia announces Sonic-3.6 TTS with major gains in naturalness across 44 languages, including realistic pauses, filler words in English, and fluid Hindi-English code-switching, released three months after Sonic-3.5 based on user feedback. 

Sonic-3.6 ranks #1 on Artificial Analysis's Provider Voice leaderboard (Elo 1282) and Controlled Voice leaderboard (Elo 1123), ahead of Speechify Simba 3.2, Alibaba's Qwen-Audio, and ElevenLabs v3 in blind tests using both native and standardized voices.

Pricing: At $49 per 1M characters via the Cartesia platform, Sonic 3.6 is the most expensive of the top three models on quality, roughly 5x Speechify AI's Simba 3.2 at $10 and nearly double Alibaba's Qwen-Audio-3.0-TTS-Plus at $27.59, though still half the price of ElevenLabs' Eleven v3 at $100

 Available immediately in beta via [cartesia.ai](http://cartesia.ai), the state-space model architecture delivers low-latency streaming performance, positioning Cartesia as a leader in production-grade, human-like voice AI.

[![User Avatar](https://storage.googleapis.com/papyrus_images/0df3909b07d27a1a63a35550f41b0eea94da50a18fb1c43898884637915c1a84.jpg)](https://twitter.com/cartesia)

[Cartesia](https://twitter.com/cartesia)

[@cartesia](https://twitter.com/cartesia)

[](https://twitter.com/cartesia/status/2089401199967559932)

Introducing Sonic-3.6: our most lifelike TTS yet, and another step change in naturalness across 44 languages. Just three months after our Sonic-3.5 launch, we've made fundamental model improvements based on feedback from the teams building on Sonic.  
  
We're #1 on [@ArtificialAnlys](https://twitter.com/ArtificialAnlys)

![](https://storage.googleapis.com/papyrus_images/f015caacccb79e73dd10e24ca605ba51afe143e5dba1cfcbc4a1dc8d461d710f.jpg)

[659](https://twitter.com/cartesia/status/2089401199967559932)[

5:17 PM • Aug 17, 2026

](https://twitter.com/cartesia/status/2089401199967559932)

[![User Avatar](https://storage.googleapis.com/papyrus_images/51db31a2ffc8cf9397bcc6fe71f8af7338664b4c3dced759d79bc3bcc2d9ec3a.jpg)](https://twitter.com/ArtificialAnlys)

[Artificial Analysis](https://twitter.com/ArtificialAnlys)

[@ArtificialAnlys](https://twitter.com/ArtificialAnlys)

[](https://twitter.com/ArtificialAnlys/status/2089400880688976062)

Cartesia's Sonic 3.6 takes the #1 spot on both the Provider Voice and Controlled Voice Artificial Analysis Speech Arena leaderboards, surpassing Speechify AI's Simba 3.2 and Alibaba's Qwen-Audio-3.0-TTS-Plus, with Sonic 3.5 holding #2 on Controlled Voice  
  
Sonic 3.6 is the latest

![](https://pbs.twimg.com/tweet_video_thumb/HP8FQfjbsAAreas.jpg)

[491](https://twitter.com/ArtificialAnlys/status/2089400880688976062)[

5:16 PM • Aug 17, 2026

](https://twitter.com/ArtificialAnlys/status/2089400880688976062)

Wednesday 19th August 2026
--------------------------

Artificial Analysis Launches Web Search API Benchmark for AI Agents

Artificial Analysis released its Search Index, evaluating web search APIs like Parallel, Exa, and Firecrawl on real agent tasks such as research and multi-hop browsing. Parallel leads with a 75 score, followed by Exa at 74 and Firecrawl at 73—all far ahead of the model-only baseline of 33. The benchmark factors in total cost and time, highlighting how top performers like Parallel and Firecrawl balance efficiency with savings on model tokens. Developers praise the independent data for picking reliable tools as AI agents lean more on web search.

[![User Avatar](https://storage.googleapis.com/papyrus_images/51db31a2ffc8cf9397bcc6fe71f8af7338664b4c3dced759d79bc3bcc2d9ec3a.jpg)](https://twitter.com/ArtificialAnlys)

[Artificial Analysis](https://twitter.com/ArtificialAnlys)

[@ArtificialAnlys](https://twitter.com/ArtificialAnlys)

[](https://twitter.com/ArtificialAnlys/status/2089755262915936661)

Announcing the Artificial Analysis Search Index, benchmarking how search API providers perform on quality, cost, and speed when used by an agent. We are initiating coverage with Parallel, Exa, Firecrawl, You (dot) com, Tavily, Keenable, and Brave  
  
Search is one of the most

![](https://storage.googleapis.com/papyrus_images/91099c8adbc5feb29030924d0f9dd87cc73dcc7b4cb39dd16bf0a7cff39239a5.jpg)

[967](https://twitter.com/ArtificialAnlys/status/2089755262915936661)[

4:44 PM • Aug 18, 2026

](https://twitter.com/ArtificialAnlys/status/2089755262915936661)

FWIW,   I wrote [this public Notion page](https://app.notion.com/p/yusuf-goolamabbas-53/Public-Benefits-of-integrating-external-search-fetch-providers-into-a-harness-3ad83c2ae30380e5aaaae89b66e20267?source=copy_link) on the benefits of integrating external search/fetch providers into a harness

K-Dense AI announced the release of a new agent skills plugin specialized in small molecule and protein therapeutics, enabling AI agents to handle tasks in cheminformatics, molecular modeling, docking, dynamics, and target discovery. 

The plugin is a curated subset of their main Scientific Agent Skills GitHub repo (K-Dense-AI/scientific-agent-skills), which has 33.8k stars, 161+ validated skills, and 100+ scientific databases covering biology, chemistry, medicine, and drug discovery 

It supports installation and use in multiple AI coding tools including Claude Code, Cursor, Codex, and Google Antigravity via the open Agent Skills standard. 

  

[![User Avatar](https://storage.googleapis.com/papyrus_images/c423217ddafff30400bcd6be63eeeb2139fbfe8a5719b3a4c5202f6cf339334a.png)](https://twitter.com/k_dense_ai)

[K-Dense](https://twitter.com/k_dense_ai)

[@k\_dense\_ai](https://twitter.com/k_dense_ai)

[](https://twitter.com/k_dense_ai/status/2089495857666400696)

We just shipped our first agent skills plugin for small molecule and protein therapeutics. This represents a curated subset of our larger Scientific Agent Skills repo (33k+ ![⭐](https://abs-0.twimg.com/emoji/v2/72x72/2b50.png) on GitHub).  
  
Try it today in Claude Code, Cursor, Codex, Google Antigravity, and more.[

![](https://storage.googleapis.com/papyrus_images/8410d74c5e7bbbee9bce8f9fc4bc5b958f94cee495095acb77d00c66699e33d1.jpg)

github.com

GitHub - K-Dense-AI/drug-discovery-agent-skills: Agent Skills for small-molecule and protein...
-----------------------------------------------------------------------------------------------

Agent Skills for small-molecule and protein therapeutics: cheminformatics, molecular ML, docking and dynamics, protein design platforms, and target-discovery knowledge graphs. - K-Dense-AI/drug-dis...





](https://t.co/bfRcHZm7DR)

[26](https://twitter.com/k_dense_ai/status/2089495857666400696)[

11:33 PM • Aug 17, 2026

](https://twitter.com/k_dense_ai/status/2089495857666400696)

Replit's new "Free Mode" lets Core and Pro subscribers use the Agent for everyday tasks (chatting, ideating, quick builds) without consuming usage credits, effectively letting them create up to 30X more within their existing monthly subscription. 

Free Mode is powered by OpenAI's GPT-5.6 Luna and applies to fast, high-value everyday tasks — chats, ideation, quick edits, and simpler builds — which no longer draw down credit-based usage.

[

Replit Introduces Free Mode | Replit
------------------------------------

AI models are now capable and affordable enough to make once-unreachable outcomes practical. The opportunity for everyone to pursue and collaborate on ambitious ideas, spend less time on busywork, and raise the quality and creative output across disciplines is nearly here. The possibility is here. But for most people, the reality is not.

https://replit.com

![Replit Introduces Free Mode | Replit](https://storage.googleapis.com/papyrus_images/9fde0ad41a7fd6d88eeaaa77c95f5dd12069c5ab7c1f40eaa046a00acc0497dc.jpg)

](https://replit.com/blog/replit-introduces-free-mode)

[Just LDA](https://t.me/JustLDA) asked the following question

Anyone here using Buzz, Grok Bot, or the new Hermes Bot function?

Opinions? I’d assume Hermes Bot works better than Grok bot because then you’re not limited to one model.

But just curious to hear from others that have used them.

I’m currently using Buzz and it seems to be working well. I have my Claude agent, Hermes agent (ChatGPT), and Grok agent all working together now on my various projects.

Thursday 20th August 2026
-------------------------

[Diamond Hands Dig](https://t.me/diamondhandsdig) responded to [Just LDA](https://t.me/JustLDA) with the following

Just wrote about Hermes Agent Bot Mode yesterday

  

[![User Avatar](https://storage.googleapis.com/papyrus_images/dce1f8989fca7b383925b6bad937e6d718ec2d270bb912ca0d536ed8a96820bb.jpg)](https://twitter.com/DiamondHandsDig)

[Diamond Hands Dig](https://twitter.com/DiamondHandsDig)

[@DiamondHandsDig](https://twitter.com/DiamondHandsDig)

[](https://twitter.com/DiamondHandsDig/status/2089483505562419221)

[x.com/i/article/2089…](https://t.co/KLUbBmeZNw)

[4](https://twitter.com/DiamondHandsDig/status/2089483505562419221)[

10:44 PM • Aug 17, 2026

](https://twitter.com/DiamondHandsDig/status/2089483505562419221)

**Google made AI Pro free for US students, AI Plus free for students in 140+ countries, and a cheap YouTube student bundle.**

Google Gemini announced a back-to-school promotion offering eligible US college students one year of Google AI Pro for free and students in 140+ countries one year of Google AI Plus at no cost, running through December 31, 2026. 

New student tools rolling out include a dedicated hub for study notebooks, flashcards, and quizzes; adaptive lessons that pinpoint knowledge gaps from class materials; and interactive 3D visualizations generated via prompts like “Show me how DNA works.” 

Gemini Live now supports hands-free multi-step research reports that run in the background, with notifications and voice discussion of findings to support deeper academic work.

[![User Avatar](https://storage.googleapis.com/papyrus_images/6ed01cb5f40ffd1310bef8f822fd5d46338638adf5d3c06c1d1908d2b6149b27.png)](https://twitter.com/GeminiApp)

[Google Gemini](https://twitter.com/GeminiApp)

[@GeminiApp](https://twitter.com/GeminiApp)

[](https://twitter.com/GeminiApp/status/2090165248196252003)

Back-to-school season is here and starting today, eligible college students can get a full year of Gemini on us:  
\- US students: 1 year of Google AI Pro at no cost  
\- 140+ countries: 1 year of Google AI Plus at no cost  
  
Here’s what’s new for students ![👇](https://abs-0.twimg.com/emoji/v2/72x72/1f447.png)

[28.5K](https://twitter.com/GeminiApp/status/2090165248196252003)[

7:53 PM • Aug 19, 2026

](https://twitter.com/GeminiApp/status/2090165248196252003)

[![User Avatar](https://storage.googleapis.com/papyrus_images/f3d53bcf816430292bf4c6cb13295ddee8c29909f4ef5127795ac5b9e00601b3.png)](https://twitter.com/StudentOffersHQ)

[Student Offers](https://twitter.com/StudentOffersHQ)

[@StudentOffersHQ](https://twitter.com/StudentOffersHQ)

[](https://twitter.com/StudentOffersHQ/status/2090245465774449147)

Google made AI Pro free for US students, AI Plus free for students in 140+ countries, and a cheap YouTube student bundle.  
  
1/ Google AI Pro — free 12 months, US only  
$240 value (was $19.99/mo)  
Gemini 3.1 Pro, 5 TB, Antigravity access  
  
2/ Google AI Plus — free 12 months,

![](https://storage.googleapis.com/papyrus_images/f44f20d61dc7945d99a0f204566f124e0f5c5e75ed186f556e23f22ef7567f4c.jpg)

[828](https://twitter.com/StudentOffersHQ/status/2090245465774449147)[

1:12 AM • Aug 20, 2026

](https://twitter.com/StudentOffersHQ/status/2090245465774449147)

[![User Avatar](https://storage.googleapis.com/papyrus_images/8d47c1bc53da10138992d55cc322f97384c815a96959a323531b1fddc049e42f.jpg)](https://twitter.com/nahid_pro09)

[Nahid](https://twitter.com/nahid_pro09)

[@nahid\_pro09](https://twitter.com/nahid_pro09)

[](https://twitter.com/nahid_pro09/status/2090274964549288431)

Free is not a typo a full year of gemini is $0 for students right now![😳](https://abs-0.twimg.com/emoji/v2/72x72/1f633.png)  
  
starting today:  
→us students: 1 year Google AI Pro, at no cost  
→140+ countries: 1 year Google AI Plus, at no cost  
  
one sign-up. zero dollars. twelve months of AI doing the heavy lifting  
  
the only thing

![](https://storage.googleapis.com/papyrus_images/d039cb7c2e3fdc70305dac562c2f0ab47e748b52b8744fc0aa10b29cc8556cc4.jpg)

[![User Avatar](https://storage.googleapis.com/papyrus_images/8d47c1bc53da10138992d55cc322f97384c815a96959a323531b1fddc049e42f.jpg)](https://twitter.com/nahid_pro09)

[Nahid](https://twitter.com/nahid_pro09)

[@nahid\_pro09](https://twitter.com/nahid_pro09)

[](https://twitter.com/nahid_pro09/status/2069884484863074794)

1 million tokens/min completely free on gemini 2.5 flash and pro - official google api, zero strings attached![😳](https://abs-0.twimg.com/emoji/v2/72x72/1f633.png)  
  
google's ai studio tier is the only developer api limit you'll ever need  
  
"genuinely giving away free ai infra" - that's the vibe here. massive context windows and

![](https://storage.googleapis.com/papyrus_images/d2396e51cb67e8cc854de6c6786b8647a33225fcdf28d0200f0b000ada951740.jpg)

[239](https://twitter.com/nahid_pro09/status/2090274964549288431)[

3:09 AM • Aug 20, 2026

](https://twitter.com/nahid_pro09/status/2090274964549288431)

Whilst not LLM related per-se,  thought this might be of interest to group members particularly those who work with Javascript/Typescript codebases which are coded/generated via their coding agents and maybe utilising nodejs or deno for their runtime

Bun 1.4 released (Claude Code has been running on a canary version of it for quite a while)

[![User Avatar](https://storage.googleapis.com/papyrus_images/2018c6354f76f49f205d07d7e14b2a44f7af3323fd6c67ab91f990fa896bb161.jpg)](https://twitter.com/bunjavascript)

[Bun](https://twitter.com/bunjavascript)

[@bunjavascript](https://twitter.com/bunjavascript)

[](https://twitter.com/bunjavascript/status/2090443675063263738)

Bun 1.4  
  
\- Fixes over 2,900 GitHub issues  
\- +1,517 tests from the Node.js test suite  
\- Reduces idle CPU by 5x  
\- Reduces memory usage by up to 35%  
\- Starts up to 50% faster on Linux  
\- Rewrites Bun in Rust  
  
Thank you everyone who contributed since Bun 1.3!  
  
[bun.com/blog/bun-v1.4](https://t.co/3YjnuhjE8s)

[4,495](https://twitter.com/bunjavascript/status/2090443675063263738)[

2:19 PM • Aug 20, 2026

](https://twitter.com/bunjavascript/status/2090443675063263738)

[

Bun 1.4
-------

Bun 1.4 rewrites Bun in Rust, ships built-in headless browser automation (Bun.WebView), Bun.Image, Bun.markdown, JSON5, JSONL, Terminal and cron APIs, Node.js 26.3.0 compatibility with 1,517 newly passing tests, parallel test and run, Windows ARM64, and an opt-in global virtual store for up to 7× faster installs.

https://bun.com

![Bun 1.4](https://storage.googleapis.com/papyrus_images/ff20703acc390d2216ab8687d0a3fc15a3ce5ae929717e03edb9f90a2ee4c194.png)

](https://bun.com/blog/bun-v1.4)

[![](https://paragraph.com/editor/youtube/play.png)](https://www.youtube.com/watch?v=i38DgEuaJwM)

The rewrite of Bun from Zig to Rust in 11 days was done with the assistance of Fable 5

[

Rewriting Bun in Rust
---------------------

Why & how we rewrote Bun from Zig to Rust

https://bun.com

![Rewriting Bun in Rust](https://storage.googleapis.com/papyrus_images/ef71a52cbd6754bd5aa1178e6d65840e4314256623ed4f94906e67b68b90e614.png)

](https://bun.com/blog/bun-in-rust)

PS:  Anthropic acquired Bun last year.  Jarred Sumner is the founder

[https://x.com/jarredsumner](https://x.com/jarredsumner)

Couple of links on the Meta AI app (which doesn't seem to have rolled out to my region, might need to VPN or Meta could have region locked based on metadata from my usage of FB/IG).   The video in Spencer Barnett's post makes it very intriguing to try it out

[![User Avatar](https://storage.googleapis.com/papyrus_images/0b55317b8c68f0a934c699b8c53bc6978f1a747575a9b8b5f7c7776bb6b7b11c.jpg)](https://twitter.com/pratanchandani)

[Prashant Ratanchandani](https://twitter.com/pratanchandani)

[@pratanchandani](https://twitter.com/pratanchandani)

[](https://twitter.com/pratanchandani/status/2090160440550842489)

Big update from us on Meta AI - this time focused on small businesses.  
  
It can now plug into your FB/IG analytics, Ads, + Google Workspace to audit campaigns, benchmark vs similar brands, and tell you what's actually working.  
  
And there's a new native Mac app — you can share any[

![](https://storage.googleapis.com/papyrus_images/c35b4a7cc2722286a411d991c61c2626c9491c336aa47ecb69e98b1d2fd5353f.png)

meta.ai

Meta AI
-------

Use Meta AI assistant to get things done, create AI-generated images for free, and get answers to any of your questions.





](https://t.co/wdQndIn80V)

[278](https://twitter.com/pratanchandani/status/2090160440550842489)[

7:34 PM • Aug 19, 2026

](https://twitter.com/pratanchandani/status/2090160440550842489)

[![User Avatar](https://storage.googleapis.com/papyrus_images/2c7c36bb2d630cfc1bf2832da87169b1c99f5a0fa01257c122f8209730f2330e.jpg)](https://twitter.com/spencerbarnett)

[Spencer Barnett](https://twitter.com/spencerbarnett)

[@spencerbarnett](https://twitter.com/spencerbarnett)

[](https://twitter.com/spencerbarnett/status/2090164939105476969)

We launched the Meta AI Mac OS app today! ![🚀](https://abs-0.twimg.com/emoji/v2/72x72/1f680.png)  
  
I particularly love the dictation feature which allows me to dictate anywhere on my computer with ultra high accuracy. Just hold down 'fn' and yap!

![](https://pbs.twimg.com/amplify_video_thumb/2090162325710159872/img/qOJyQ9SeQzQ1ullK.jpg)

[791](https://twitter.com/spencerbarnett/status/2090164939105476969)[

7:52 PM • Aug 19, 2026

](https://twitter.com/spencerbarnett/status/2090164939105476969)

[![User Avatar](https://storage.googleapis.com/papyrus_images/0ba931c49f15868e9153fad0df159edcbe7ce220f21beb102cedf64bd5407362.jpg)](https://twitter.com/alexandr_wang)

[Alexandr Wang](https://twitter.com/alexandr_wang)

[@alexandr\_wang](https://twitter.com/alexandr_wang)

[](https://twitter.com/alexandr_wang/status/2090289743129194523)

check out the Meta AI desktop app!  
  
the dictation has been a game changer for me personally

[![User Avatar](https://storage.googleapis.com/papyrus_images/2c7c36bb2d630cfc1bf2832da87169b1c99f5a0fa01257c122f8209730f2330e.jpg)](https://twitter.com/spencerbarnett)

[Spencer Barnett](https://twitter.com/spencerbarnett)

[@spencerbarnett](https://twitter.com/spencerbarnett)

[](https://twitter.com/spencerbarnett/status/2090164939105476969)

We launched the Meta AI Mac OS app today! ![🚀](https://abs-0.twimg.com/emoji/v2/72x72/1f680.png)  
  
I particularly love the dictation feature which allows me to dictate anywhere on my computer with ultra high accuracy. Just hold down 'fn' and yap!

![](https://pbs.twimg.com/amplify_video_thumb/2090162325710159872/img/qOJyQ9SeQzQ1ullK.jpg)

[1,261](https://twitter.com/alexandr_wang/status/2090289743129194523)[

4:08 AM • Aug 20, 2026

](https://twitter.com/alexandr_wang/status/2090289743129194523)

Friday 21st August 2026
-----------------------

DeepSeek Launches Experimental Vision Model Matching Top Benchmarks

The new DeepSeek-V4-Flash-Vision-Exp adds image-handling to the V4-Flash series, scoring big on multimodal benchmarks like 36.5 on ApexBench and 27.3 on Agents' Last Exam—often rivaling Opus-4.8—while holding steady on pure-text tasks. Developers access it easily via the DeepSeek API with base64 images, URLs, or the fresh Files API, all at standard pricing. Community voices, from researchers to tech analysts, praise the leap in visual agent performance without speed or cost hits, signaling a push toward versatile multimodal tools.

[![User Avatar](https://storage.googleapis.com/papyrus_images/5e1311e1bfa0da893f7934d2831628d79c254fbc3e39a50634635be0803c483e.jpg)](https://twitter.com/deepseek_ai)

[DeepSeek](https://twitter.com/deepseek_ai)

[@deepseek\_ai](https://twitter.com/deepseek_ai)

[](https://twitter.com/deepseek_ai/status/2090730032574631962)

DeepSeek-V4-Flash-Vision-Exp is now live on the DeepSeek API Platform! ![🚀](https://abs-0.twimg.com/emoji/v2/72x72/1f680.png)  
  
![🔹](https://abs-0.twimg.com/emoji/v2/72x72/1f539.png) This experimental multimodal model matches DeepSeek-V4-Flash on text capabilities—including agents, reasoning, and world knowledge.  
![🔹](https://abs-0.twimg.com/emoji/v2/72x72/1f539.png) On multimodal agent benchmarks, V4-Flash-Vision-Exp makes a major

![](https://storage.googleapis.com/papyrus_images/91f795fba7d9750127b802e87615c0f21a4a082922fd1116007b39fa0f0395c3.jpg)

[5,733](https://twitter.com/deepseek_ai/status/2090730032574631962)[

9:17 AM • Aug 21, 2026

](https://twitter.com/deepseek_ai/status/2090730032574631962)

[![User Avatar](https://storage.googleapis.com/papyrus_images/682a8e88fc2049c99b9d357474becf6ad383153036b545e0fbacec532398ebd9.jpg)](https://twitter.com/kimmonismus)

[Chubby♨️](https://twitter.com/kimmonismus)

[@kimmonismus](https://twitter.com/kimmonismus)

[](https://twitter.com/kimmonismus/status/2090738285140173088)

wtf is happening today: DeepSeek has launched DeepSeek-V4-Flash-Vision-Exp, an experimental multimodal model built for agents that need to see.  
  
And its performance on visual-agent benchmarks moves close to or even outperforms Opus 4.8.  
  
Again: this is the Flash model, the

![](https://storage.googleapis.com/papyrus_images/8679b991acf461851f1c69f6899039ee504c679acaadbd53e6885f78af21182c.jpg)

[![User Avatar](https://storage.googleapis.com/papyrus_images/5e1311e1bfa0da893f7934d2831628d79c254fbc3e39a50634635be0803c483e.jpg)](https://twitter.com/deepseek_ai)

[DeepSeek](https://twitter.com/deepseek_ai)

[@deepseek\_ai](https://twitter.com/deepseek_ai)

[](https://twitter.com/deepseek_ai/status/2090730032574631962)

DeepSeek-V4-Flash-Vision-Exp is now live on the DeepSeek API Platform! ![🚀](https://abs-0.twimg.com/emoji/v2/72x72/1f680.png)  
  
![🔹](https://abs-0.twimg.com/emoji/v2/72x72/1f539.png) This experimental multimodal model matches DeepSeek-V4-Flash on text capabilities—including agents, reasoning, and world knowledge.  
![🔹](https://abs-0.twimg.com/emoji/v2/72x72/1f539.png) On multimodal agent benchmarks, V4-Flash-Vision-Exp makes a major

![](https://storage.googleapis.com/papyrus_images/91f795fba7d9750127b802e87615c0f21a4a082922fd1116007b39fa0f0395c3.jpg)

[860](https://twitter.com/kimmonismus/status/2090738285140173088)[

9:50 AM • Aug 21, 2026

](https://twitter.com/kimmonismus/status/2090738285140173088)

Mysterious Ox Alpha AI Model Debuts with Strong Coding Prowess

Ox Alpha, listed on OpenRouter as a free preview from an anonymous provider, handles 1 million token contexts with text, image, and video inputs, excelling in coding and agentic tasks. Early tests by developer Ben Davis showed it hitting 80% on DeepSWE benchmarks, outpacing models like Fable at 65% and gpt-5.6-sol at 52%. Community speculation points to Chinese origins, possibly Zhipu’s next GLM or Xiaomi’s MiMo, with its polished speed and frontend strengths fueling excitement during the limited free access period.

[![User Avatar](https://storage.googleapis.com/papyrus_images/fcb57ac531b7bcea96bede3d02b1c04020bcbc6e11aa5582f8cc8c19d7077098.jpg)](https://twitter.com/OpenRouter)

[OpenRouter](https://twitter.com/OpenRouter)

[@OpenRouter](https://twitter.com/OpenRouter)

[](https://twitter.com/OpenRouter/status/2090544970923184269)

![🥷](https://abs-0.twimg.com/emoji/v2/72x72/1f977.png) New stealth model: Ox Alpha  
  
Ox Alpha is a frontier model built for efficient coding, sustained agentic work, and real-world production use.  
  
\- 1M token context window  
\- Text, image, and video input  
  
Try it now and share feedback to improve the model![

![](https://storage.googleapis.com/papyrus_images/3e01661533bde18d488cb36ae31f31c597f2c5952d23b49d00e484e19ada5591.jpg)

openrouter.ai

Ox Alpha - API Pricing & Providers
----------------------------------

Ox Alpha is a reasoning model designed for coding, sustained agentic work, and production workloads. This model is free to use. 1,048,576 token context window, maximum output of 131,072 tokens.





](https://t.co/tU5lmZrO5z)

[3,604](https://twitter.com/OpenRouter/status/2090544970923184269)[

9:02 PM • Aug 20, 2026

](https://twitter.com/OpenRouter/status/2090544970923184269)

[![User Avatar](https://storage.googleapis.com/papyrus_images/920a3a4dc1c89e745da64c6e1e9a3d9c802e7b40c8090634a005d862c32ec3a3.jpg)](https://twitter.com/CommandCodeAI)

[Command Code](https://twitter.com/CommandCodeAI)

[@CommandCodeAI](https://twitter.com/CommandCodeAI)

[](https://twitter.com/CommandCodeAI/status/2090716833578090593)

Ox Alpha (stealth frontier model) is free in Command Code.  
  
![🆓](https://abs-0.twimg.com/emoji/v2/72x72/1f193.png) 100T tokens per day free capacity, no 5hr usage limits.  
  
• 1M content  
• we benchmarked ~5B tokens  
• built a tool repair harness to imp tool calling  
  
absolutely free for all plans.  
  
guess who made it?  
  
![🐐](https://abs-0.twimg.com/emoji/v2/72x72/1f410.png)

[2,029](https://twitter.com/CommandCodeAI/status/2090716833578090593)[

8:25 AM • Aug 21, 2026

](https://twitter.com/CommandCodeAI/status/2090716833578090593)

[![User Avatar](https://storage.googleapis.com/papyrus_images/996d2635b565461ab4c660268cbab5f26705efe7f3be324041d55100b39288d9.png)](https://twitter.com/opencode)

[OpenCode](https://twitter.com/opencode)

[@opencode](https://twitter.com/opencode)

[](https://twitter.com/opencode/status/2090544355824038300)

Ox Alpha (stealth model) is free for the next week  
  
\- 1M Context  
\- Multi-modal  
\- Zero Data Retention  
  
Generous rate limits, near unlimited usage  
  
We have capacity for 100T tokens per day, lets see what you can do

[13.5K](https://twitter.com/opencode/status/2090544355824038300)[

8:59 PM • Aug 20, 2026

](https://twitter.com/opencode/status/2090544355824038300)

[![User Avatar](https://storage.googleapis.com/papyrus_images/682a8e88fc2049c99b9d357474becf6ad383153036b545e0fbacec532398ebd9.jpg)](https://twitter.com/kimmonismus)

[Chubby♨️](https://twitter.com/kimmonismus)

[@kimmonismus](https://twitter.com/kimmonismus)

[](https://twitter.com/kimmonismus/status/2090718270202528215)

Wtf! Ben ran this mystery model through 10 DeepSWE tasks and it scored over 80%, versus 65% for Fable and 52% for GPT-5.6-sol!  
  
this is insane. Probably a chinese company. Either a new GLM or Kimi model, I reckon.

[![User Avatar](https://storage.googleapis.com/papyrus_images/d3e3869179bdb319416ae921c5bce850d48c20a0f373e72c8e5ec61a6b1dd335.jpg)](https://twitter.com/davis7)

[Ben Davis](https://twitter.com/davis7)

[@davis7](https://twitter.com/davis7)

[](https://twitter.com/davis7/status/2090655207831298095)

I ran this thing through 10 tasks on DeepSWE (so there could be a ton of variance in it's real score, this is a subset), but uh...  
  
gpt-5.6-sol: 52%  
fable: 65%  
whatever the hell this is: 80% (was a near miss on the "x"s so actually over 80%)  
  
I am very confused

![](https://storage.googleapis.com/papyrus_images/6171df7f06fd4bce48f719a86ba48d92356fa78b917ae3908ca594933b280319.png)

[1,799](https://twitter.com/kimmonismus/status/2090718270202528215)[

8:30 AM • Aug 21, 2026

](https://twitter.com/kimmonismus/status/2090718270202528215)

Saturday 22nd August 2026
-------------------------

* * *

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 Seedream 5 Pro model within the [**Krea**](https://www.krea.ai/refer/EJQQP9FJ) tool via the following prompt

![](https://paragraph.com/editor/callout/information-icon.png)

An enormous floating city in the sky, steampunk style with lots of towers and domes, oil painting, fantasy art, detailed

---

*Originally published on [This Week in All Things AI](https://paragraph.com/@twiata/this-week-in-all-things-ai-week-34-2026)*
