Omkar Chebale

Hi friends! 👋 I love simplifying complex tech topics into easy-to-understand, beginner-friendly blogs. Want curated content tailored just for YOU? Subscribe to my newsletter now! 🔗 omkarchebale.vercel.app/blog

Speculative Decoding — The Intern Trick That Makes LLMs 13x Faster omkarchebale.vercel.app/blog/speculati…

Excited to share that I’ve officially joined AI Planet as an AI Engineer! 🚀 When I first

Been using AI since ChatGPT first came out. My reading speed, my research, the way I analyze and search, even how I prompt… all of it got faster and sharper. But the real shift? It’s how much more involved I feel now. If you use AI every day, you already know what I mean.

Back to reading

@Shreyas_Pandeyy @mynk8s @MiniMaxAgent @MiniMax_AI What u gonna build ?

@mynk8s @MiniMaxAgent @MiniMax_AI are bhai pata hai kuch credits mil jaate toh kuch tagda bana leta

hey, anyone from team @MiniMaxAgent @MiniMax_AI can i get some credits to try the new M3?

@mynk8s @Shreyas_Pandeyy @MiniMaxAgent @MiniMax_AI Not anymore

@Shreyas_Pandeyy @MiniMaxAgent @MiniMax_AI I think it's free on opencode for now

10 GitHub repos so good they shouldn't be free. 1. AutoHedge An autonomous hedge fund built in Python with four AI agents: a director generates investment theses, a quant validates them, a risk manager decides position size, and an execution agent places orders. Operates live on Solana. With 'pip install -U autohedge', you can start trading immediately. repo → github.com/The-Swarm-Corp… 2. Vibe-Trading A trading system using a Directed Acyclic Graph (DAG) model, featuring 64 finance skills and 29 preset specialist agent swarms. Includes analysis methods like Ichimoku, Elliott Wave, SMC, Black-Scholes, full Greeks, and risk parity. Its crypto desk provides liquidation heatmaps and token unlock tracking. You can observe agents debating strategies in real time. repo → github.com/HKUDS/Vibe-Tra… 3. Fincept Terminal A Bloomberg Terminal replacement that runs on your laptop. CFA levels 1, 2, and 3 analytics. 20+ investor AI agents (Buffett, Dalio, Soros). 100+ data connectors, including Polygon, World Bank, and IMF. Bloomberg charges $24,000 a year. This is free. repo → github.com/Fincept-Corpor… 4. LibreChat Every model ChatGPT runs, plus Claude, Gemini, DeepSeek, and 20 more. Self-hosted. Native MCP support. You own the data, the history, the infrastructure. OpenAI charges $20/month to use their wrapper. This costs nothing to use your own. repo → librechat.ai 5. Open Higgsfield AI A self-hosted cinema studio with 200+ AI models. Flux, Midjourney, Sora, Kling, Veo, GPT-4o, SDXL all in one interface. Text to image. Image to video. Cinema mode with pro camera controls. No subscription. Your data stays local. repo → github.com/Anil-matcha/Op… 6. Open-LLM-VTuber A Live2D AI companion that runs offline, sees your screen, hears your voice, and never forgets. Inner thoughts are shown as a separate text layer, so you watch the reasoning happen before words come out. Pet mode floats it on your desktop. Swap the LLM in one config line. repo → github.com/Open-LLM-VTube… 7. Claude Ads A free Claude Code skill that runs 190 audit checks across Google, Meta, YouTube, LinkedIn, TikTok, and Microsoft Ads. 6 parallel subagents firing at once. Consolidates into a single Ads Health Score ranked by revenue impact. Agencies charge $4,000 a month for this. repo → github.com/AgriciDaniel/c… 8. Agentic Inbox Cloudflare just open-sourced an email client where an AI agent reads your inbox and drafts your replies. Runs entirely on Cloudflare Workers. Each mailbox lives in its own Durable Object. Your email never leaves your Cloudflare account. One click deploys it. repo → github.com/cloudflare/age… 9. Camofox Browser An open source headless browser that makes AI agents invisible to bot detection. Spoofs navigator properties, WebGL, AudioContext, and WebRTC at the C++ level. The browser does not look modified because it genuinely is not. Accessibility tree output drops token cost by 90%. repo → github.com/jo-inc/camofox… 10. Hyperframes HeyGen open-sourced a video framework that does everything Remotion does without React, without JSX, without teaching your AI agent a new format. The agent writes HTML. The framework renders MP4. GSAP, Lottie, and Three.js all work. Same HTML always produces the same file. repo → github.com/heygen-com/hyp… These are not toys. Each one replaces a paid product you're still being charged for. Pick one. Install it. Plug it into your workflow. 100% free. 100% open source.

@mattpocockuk That’s why I’m typing “work flow”

So every time I say the word 'workflow' in Claude Code... (let's say, when I'm creating a new GitHub workflow) ...it tries to enter 'workflow' mode, spinning up dozens of subagents to complete my task. Stupid fucking thing

Prefill & Decode phase model :

Confession: I was on @WisprFlow for 6 months, kept hitting the free limit, and was juggling multiple accounts to keep going. Just switched back to @openwhispr. Barely any difference. Free Grok API key = ~2000 req/day, zero cost. Open source. Your key, your data.

@kirat_tw what’s new in it. researched about it, feels like one more another terminal agents.

Devin cli is something. Feels snappier than the rest.

@HarryAl16842685 @warpdotdev Thanks Harry! Already submitted PR #50 with the fix — native Windows toasts as a fallback, 16 tests, zero impact on macOS/Linux. Also shipped a standalone plugin that works with any terminal: github.com/Chebaleomkar/c… Happy to iterate if the team has feedback.

@chebalerushi @warpdotdev Thanks for flagging this @chebalerushi. Unfortunately, the claude code plugin in general was broken by a claude code update that disabled plugins from accessing the terminal. They released a work around though, and we're planning on releasing a holistic fix soon!

Fixed desktop notifications for @warpdotdev claude Code plugin on Windows 🔧 The problem: OSC 777 sequences get swallowed by Claude Code's sandbox on Windows — /dev/tty doesn't work. The fix: native Windows toasts via PowerShell, branded as Warp, with deduplication.

I analyzed 45 days of my Claude Code conversations and found I was running 50+ git commands per session doing the same thing. So I built slash commands that do it in one click. Here's exactly how (with a prompt you can steal): omkarchebale.vercel.app/blogs/how-i-au…

Instead of JSON, if diagnostic could be in TOON , then that could be token-efficient as well. @vercel

@sureshkanbu @warpdotdev Exactly — the silent 30 minutes is the worst part. You don't even know you're missing notifications until you switch back and see Claude finished 5 minutes ago 😅 Hoping @warpdotdev picks up the upstream issue so this works out of the box for Windows users.

@warpdotdev Full writeup + code → omkarchebale.vercel.app/blogs/how-i-fi… Issue filed → github.com/warpdotdev/cla… @waborern @AnthropicAI

What's stopping you from working like this?

Totally agree 👍 with you Sir .

@devendrasm @svembu What you use for persistence context ?

@svembu @svembu Ji, the cost will come down over time. Smaller models are more capable now. 1m context is not needed for most coding. Open models are similarly capable now

Important post from Meta engineer Arnav Gupta on all the AI-led layoffs. "The layoffs will continue till we learn to use AI" is his title and Gemini(!) correctly identified it as coming from, "The beatings will continue until morale improves". x.com/i/status/20518… As he explains well, AI has increased costs massively for all tech companies. Our own AI bill is skyrocketing and to add insult to injury, server prices have gone up 200-300% in a year because the AI infrastructure boom is consuming all the advanced memory chips. So these layoffs are the economic response by tech companies to control the main cost they control (people) to pay for AI and servers. Of course most are spinning it as the result of the "AI productivity miracle" but reality is more cost control than a productivity miracle, at least not yet. Now let me go back to using AI to generate even more code, so that we don't fall behind all the other guys generating massive amounts of code 😅