spencer

@ManasKushwaha99 @llmdevguy Not anymore it doesn’t

@llmdevguy Qwen CLI provides more for free, 30,000 requests per month. How is this good?

🔥Around 50500 requests per month!! Amazing value for just 5$ for first month!! This is great model, I love it for design work!

@sunnova_downy Looks great, love the design

無料です。アプリを作りました。 不眠症とADHDで、夜になると思考が止まらなくなる。 既存のアンビエントアプリを色々試したけど、どれも数分でループに気づいて余計に眠れなくなった。だから作りました。 ループも録音も使わないリアルタイム音響合成のアンビエントアプリ。 apps.apple.com/jp/app/atmós/i…

@itsjoaki 3.6 plus is working incredibly well for me in existing large codebases. Context fetching seems great and haven’t had it hallucinate or use deprecated methods.

GLM 5.1 or Qwen 3.5?

@ibocodes Pro was $30 yesterday lol

chinese models used to be the cheap alternative why are they priced like claude & codex now 😭

@mark_k Same except Qwen for code, it just aligns more with my way of programming.

My current AI setup: - SuperGrok for all daily tasks. Research, writing, assistance. It's amazing with Grok 4.20 - Codex for coding. The best coding AI currently, and the desktop app is great. These two cover all of my needs. I just unsubscribed Gemini. What's your setup?

@TeksEdge I’ve been so pleasantly surprised with qwen 3.6 is my daily driver for coding. I just always end up stuck wrestling the model into oblivion at some point, qwen has been so much better in this regard

💡 If you're tracking LLM hallucination benchmarks, (which is the most important IMHO) 📊 Qwen3.6-Plus is the best accoridng the BridgeBench!

@Alibaba_Qwen Awesome updates qwen 3.6 plus in this cli harness is so good at understanding my intentions

🚀 Qwen Code v0.14.0 – v0.14.2 are now available Channels：Control Qwen Code remotely from Telegram, DingTalk, or WeChat — send a message from your phone, get results on your server Cron Jobs ：Schedule recurring AI tasks — auto-run tests every 30 min, pull & build every morning, monitor logs on a timer Qwen3.6-Plus ：New flagship model with 1M token context, 1,000 free daily requests Sub-agent Model Selection：Assign different models to sub-agents — use a powerful model for the main task, a fast one for subtasks, save tokens without sacrificing quality /plan：Enter planning mode before execution — AI maps out all files and steps first, you confirm, then it executes Follow-up Suggestions：AI suggests 2-3 next steps after completing a task — "Add unit tests?" "Check similar files?" — click to continue Adaptive Output Tokens ：Default 8K output, auto-escalates to 64K when truncated — no more manually tuning max_tokens Ctrl+O Verbosity Toggle ：Switch between verbose and compact output mid-conversation — debug mode when you need it, clean mode when you don't 📋 Full changelog: github.com/QwenLM/qwen-co…

@aaronp613 They have done this countless times before

It appears Apple has started to push out app updates on behalf of third party apps “This update from Apple will improve the functionality of this app. No new features are included”

@ShitpostRock2 I fall for that bastard every time

I love how Todd Howard’s entire reputation is being a complete liar and snake oil salesman, and people still somehow trust him every single time

@xDaily I’m not sure he’s making the point he wants to by the end of it.

NEWS: British MP Max Wilkinson complains to a panel that X allows users to criticize immigration that will spread "not in a way I'm comfortable with". Many X users in the UK have praised Elon's ownership of X for preserving an important outlet for free speech in the country.

@akt206 @MicrosoftEdge @brave The whole menu is beyond broken. I just use @raycast on windows.

broo @MicrosoftEdge @brave

@louszbd Will the plans ever carry over to faster priority assess to the web chat?

finally glm-5.1 at the very beginning we were teaching models how to write code, basically training a system that could imitate developers. back then AI lived inside the IDE as an intelligent assistant, but we were still the main driver. that was the copilot era of AI coding. then it started to become something more collaborative. we could express a vague intention (prompt), and the model translates that intention into structured software. in a way, that was the first time we taught machines to understand vibe. earlier this year, we entered the agentic engineering era. we stopped programming line by line. models began to form plans, maintain them, and operate inside a feedback loop. the model takes responsibility for planning. and now we are approaching a moment where AI can operate on the same time horizon as engineers. this is why we built glm-5.1. we want to unlock a new long-horizon paradigm. where it starts to tackle the kinds of problems that unfold over weeks: debugging, integration. an agent to remember context over long stretches, still stay aligned with the objective (and keep correcting itself along the way)

@Designarena @xai Great stuff, now give us an official cli

BREAKING: Grok 4.20 Reasoning by @xai is #2 on Web App Arena, with an Elo of 1335. Web App Arena evaluates real-world performance on building multi-file React web apps from start to finish. Huge congrats to the @xai team for this achievement!

@MiniMax_AI Would be awesome to get some agent usage bundled in, a one plan covers all deal.

Introducing the MiniMax Token Plan: First All-Modality API Subscription Flat-rate API access to MiniMax's leading text, speech, music, video, and image models. Stop juggling multiple unpredictable bills for different modalities. One key. One predictable bill. All modalities.

@bygregorr @TechCrunch What are you on about

@TechCrunch I'm not sure why this is treated as a gotcha. Building on top of existing models is literally how the whole AI stack works right now. OpenAI, Anthropic, everyone layers. The interesting question is what Cursor added on top.

Cursor admits its new coding model was built on top of Moonshot AI’s Kimi techcrunch.com/2026/03/22/cur…

@robinebers Unless what happens in the post you referenced

opensource models have zero chance to win imo the things they do well will be copied into commercial closed-source models they will never be able to catch up they will survive, but they cannot win

@fynnso Doesn’t it need a powered by Kimi notice somewhere?

was messing with the OpenAI base URL in Cursor and caught this accounts/anysphere/models/kimi-k2p5-rl-0317-s515-fast so composer 2 is just Kimi K2.5 with RL at least rename the model ID

@nish_desai @zats It is, this is just a waste of time

@zats Wait I thought the 2x limits were applied to the CLI as well!? Oops!

How to get 2x limits on codex cli through April 2 🫢 $ 𝚌𝚊𝚛𝚐𝚘 𝚋𝚞𝚒𝚕𝚍 --𝚛𝚎𝚕𝚎𝚊𝚜𝚎 -𝚙 𝚌𝚘𝚍𝚎𝚡-𝚌𝚕𝚒 --𝚋𝚒𝚗 𝚌𝚘𝚍𝚎𝚡

@bluesky Sweet, see ya

Some platforms optimize for getting you to waste hours scrolling. We're actually fine with it if you close Bluesky and live your life

@homemadehooplah @sliceoflife1311 The tool using you? People really are stupid

@sliceoflife1311 Every previous creative tool still required talent, effort, or creativity to use it. Writing a prompt and having a machine do all the work is not you using the tool, it's the tool using you.

AI is a creative shortcut that no one is ever going to respect.