←
God of Prompt1,401 posts
God of Prompt avatar

God of Prompt

@godofprompt

🔑 Your key to AI independence Get powerful prompts, custom workflows, guides and essential info on all you may need in AI

◷ Joined June 2026▧ 205 Photos and videos
God of Prompt avatar
God of Prompt@godofprompt·2h

You pay $20 a month for ChatGPT Plus. Since yesterday that includes GPT-6.2 Astra, OpenAI's strongest model, at less than a third of the cost per task of the last Astra. OpenAI says you'll see the difference most in slideshows, websites and 3D designs. So skip "make me a deck about X" and give it the brief a designer would ask you for. What changed for you: → GPT-6.2 Astra and Sol are in ChatGPT Work and Codex on Plus, Pro, Enterprise and Edu → Pro 200 gets Ultrafast mode for the first time, starting with GPT-6.2 Sol → Tighter safeguards on high-risk cybersecurity requests The prompt I'd run first: "Build a 10-slide deck on [topic] for [audience] who must decide [decision] by the end. Slide 1 states the decision and my recommendation in one sentence. Every slide after it has one claim, one number or example that proves it, and a headline that states the takeaway. Use [brand colors / plain black and white]. No stock photos and no slide over 30 words. Last slide: the decision, the cost, and the next step with an owner and a date. Before you build, list every number you'd need from me instead of guessing it." Run it once on Astra and once on the model you used last week. The gap is your review. We keep prompts like this sorted by task in the God of Prompt library, free. Link below. What would you test first: a deck, a site or a 3D design?

OpenAI@OpenAINov 17

GPT-6.2 Astra—a frontier model built for your hardest projects. It sets new highs on evals across professional work and scientific research at less than a third of the cost per task of GPT‑6 Astra. You’ll see the difference in things like slideshows, websites, and 3D designs.

Post attachmentPost attachmentPost attachment
God of Prompt avatar
God of Prompt@godofprompt·9h

You're using ChatGPT like an intern. "Analyze this." "Make a plan." "Give me ideas." Then the answers read like an intern wrote them. 7 prompts that make it work like a senior hire. Copy, paste, fill the brackets. Prompt 1: Ask before answering "Before you answer, tell me what you think I'm actually trying to solve and ask me the 3 questions that would change your answer most. Wait for my replies." Prompt 2: Pre-mortem "It's 6 months from now and [plan] failed. List the 5 most likely reasons, ranked by probability, and one early warning sign for each that I can check this week." Prompt 3: Decision table "I'm choosing between [A] and [B]. Build a table: cost, time, risk, and what I give up by not picking the other. Then tell me which you'd pick and what would change your mind." Prompt 4: A plan that fits my week "Turn [goal] into a 7-day plan using my real limits: [hours per day, budget, tools]. One task a day, under 90 minutes each, with what done looks like." Prompt 5: Edit, don't rewrite "Edit this for clarity without changing my voice: [text]. Show only the changes as before and after, and explain each in under 10 words." Prompt 6: Ideas with a filter "Give me 20 ideas for [goal], then cut to the 3 cheapest to test this week. For each: the first step and how I'll know it worked." Prompt 7: The toughest reader "Here's my draft: [paste]. You're the toughest reader in [audience]. Where do you stop reading, and what did you expect to see that isn't there?" Save the 2 you'll actually use. The other 7,200+ are in the God of Prompt library, free, no signup. Link in the first reply. Which of these do you already do without a prompt?

Post attachment
God of Prompt avatar
God of Prompt@godofprompt·Nov 17

You describe your product as "high quality" and "fast shipping". So do all your competitors. Nobody buys a claim they've read 40 times. They buy a fact they can check. These 3 prompts find the 3 selling points you can actually prove, in about 15 minutes, with the AI you already pay for. Run them in order, in one chat. Paste your product page, specs, pricing, reviews and anything you know about competitors. Prompt 1: Fact inventory "You are a product analyst. From the materials below, list every fact about the product: specs, features, prices with their conditions, guarantees, claims. Put each in a table with its source and an evidence status: verified (test, certification, independent data), brand claim (we say so), or missing evidence. Mark words like 'best' or 'leading' as brand claims. Never invent a fact, a number or a competitor comparison. End with a list of what's missing. Materials: [paste]" Prompt 2: Selling points from facts only "Using only the verified facts from the table, find up to 3 selling points. For each one give me: the fact, the job the customer is trying to get done, the result they get, what we're comparing against, and one line of messaging. Drop anything that rests on a brand claim. If you have no competitor data, don't use 'unique' or '#1'." Prompt 3: Pick the lead "Rank these selling points. Put first the one that removes the biggest reason people don't buy. Real customer data beats our opinion. If two are too close to call, keep both and tell me the one test that would decide it: a lost-sale call, a split test or a competitor check." What you get: a table that shows which of your claims you can stand behind, 3 selling points built on proof, and the one to lead your page, ad or pitch with. A page with three facts people can check beats a page with ten promises nobody can. All 3 prompts are in the God of Prompt library along with 5 more selling-point prompts we added this week, with fill-in fields, free, no signup. Link in the first reply. Which claim on your own page could you actually prove today?

Post attachment
God of Prompt avatar
God of Prompt@godofprompt·Nov 17

Finally! the most professional model out there

Post attachment
OpenAI@OpenAINov 17

Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…

1:59
God of Prompt avatar
God of Prompt@godofprompt·Nov 17

You saved a library of AI prompts this week. You'll use maybe 3 of them. 14,000 people bookmarked one free prompt library in a single day, and half a million saw it. Bookmarks are where prompts go to die. Here's how to turn any prompt you saved into one that works on your actual task. Why saved prompts let you down: someone wrote that prompt for their task, so it's missing what yours needs. Who the result is for, what you already have, and what done looks like. The model fills those gaps with averages, and you get the average answer. You don't need a longer list. You need 2 minutes to adapt the one you picked. Prompt 1: Adapt this prompt "Here is a prompt I found: [paste]. Here is my actual task: [one sentence]. Before you rewrite it, ask me up to 5 questions about what's missing: who the output is for, what I already have, my limits (length, tone, budget, deadline), and what a finished result looks like. Then rewrite the prompt for my task. Keep what works in the original, cut anything that doesn't apply to me, and end with one line that says exactly what done looks like. Show me only the new prompt." Prompt 2: Test it before you trust it "Run the prompt above on my task. Grade your answer against the done line from 1 to 10 and list what's missing. Rewrite the prompt once to fix the biggest gap and run it again. Stop after 2 rounds and show me both answers side by side." Then save the adapted version, not the original. That's the one you'll actually open again. A list of prompts is a starting point. The prompt that works is the one that knows your task. The God of Prompt library is built for that second step: 7,200+ prompts with fill-in fields for your own context, free, no signup. Link in the first reply. How many prompts have you saved that you never opened again?

Post attachment
God of Prompt avatar
God of Prompt@godofprompt·Nov 16

You pay twice as much per token for Claude Fable 5.2. On some tasks it still costs you less than Opus 5.5. On others you pay 2.8x more for the same result. Most people switch to the newest model and watch their limit melt. Here's how to tell which task needs which model before you hit enter. The two numbers everyone is arguing about: → Anthropic: on its own coding test, Fable 5.2 scored the same as Opus 5.5 at max effort (45.3% vs 45.5%) for about half the cost per task → Garry Tan, CEO of Y Combinator: on 52 coding tasks in Claude Code, Fable solved about as many as Opus (82.7% vs 84.6%) and cost 2.8x as much per solved task. 142K people saw it Both are true. Anthropic compared Fable with Opus turned all the way up to max effort, its most expensive setting. Most people run Opus on the default, where it costs far less. And Tan says it himself: Opus already solves most everyday tasks, so Fable only pulls ahead on the hard ones. That's where it earns its price. Everywhere else you pay double for the same answer. The rule: Opus 5.5 by default, Fable 5.2 only when the task proves it needs it. Two prompts to make that automatic. Prompt 1: Route this task (paste before any big task) "You are my model router. I use Claude Opus 5.5 by default and Claude Fable 5.2 only when it pays off, because Fable costs twice as much per token. Here is the task: [describe it]. Score it on 5 signals, yes or no: 1) more than 5 steps that depend on each other, 2) it needs many files or a long document held at once, 3) a wrong answer is expensive to fix, 4) Opus already failed this or something like it, 5) there is no quick way to check the result. Recommend Fable only if 2 or more are yes. Give me the scores, the recommendation in one line, and a one-paragraph brief for that model that says what done looks like. If the task is vague, ask me 2 questions instead of guessing." Prompt 2: Rule for your Claude Code or project instructions "Default model: Opus 5.5. Before any task, check these 5 signals: dependent steps over 5, many files or long documents, expensive mistakes, earlier failures on Opus, no quick way to verify. If 2 or more apply, tell me before you start: 'This looks like a Fable task because [signals]. Switch?' If a task fails twice, suggest Fable once and say why. If I'm already on Fable and a task has 0 or 1 signals, tell me Opus can handle it." That's Fable's results on the tasks that need it, and your weekly limit kept for everything else. Our free Claude Fable 5.2 Mastery Guide goes deeper: the F.A.B.L.E. brief, when Fable pays off and when Opus is enough, real costs and fast mode, 10 briefs built around what changed in 5.2. Link in the first reply. Which model is your default right now, and did your limit notice?

Post attachment
God of Prompt avatar
God of Prompt@godofprompt·Nov 16

Stripe reinforces the monopoly

Lemon Squeezy is winding down. Over the last two years, we’ve taken everything we learned and built Managed Payments, Stripe’s merchant of record solution. We care about supporting as many businesses as possible. We also want to offer an excellent product and rapidly ship improvements: increasingly, the best way to do so is by offering a single product instead of running two platforms. Lemon Squeezy closed new store sign-ups today and we’ll stop processing payments on March 31, 2027. We know this is disruptive. We’re reaching out to all businesses using Lemon Squeezy and sharing next steps specific to their business. Our support team is here to help throughout the transition. To every founder, developer, and creator who built with us, thank you. lemonsqueezy.com/blog/saying….

God of Prompt avatar
God of Prompt@godofprompt·Nov 16

You have a 2-hour recording you'll never cut. A call, a webinar, a podcast. Grok Bot just turned a 23-hour SpaceX livestream into a 1-minute highlight reel, and Elon shared it to 1 million views. You don't need Grok Bot for the hard part. The hard part is picking the moments, and any AI does that from a transcript. Here's how. What Grok Bot did: downloaded the whole stream, picked the best moments, cut them together, added effects and music. The top question under the post: how much usage did that take? The part worth stealing is the selection. Cutting 5 clips takes 15 minutes in any app. Finding those 5 clips in 2 hours of talk is what kills your afternoon. This works for talk. For rocket footage you still need your own eyes. How to do it on the plan you already pay for: → Get the transcript with timestamps (Zoom, YouTube and Riverside all give you one) → Paste it into Claude, ChatGPT or Gemini with Prompt 1 → Cut the timestamps it gives you. CapCut or your phone's editor is enough → Use Prompt 2 to write the post for each clip Prompt 1: Find the moments "You are a video editor who cuts long recordings into short clips people finish and share. Here is a transcript with timestamps: [paste]. The audience is [who will watch]. Find the 5 best moments for clips of 30 to 60 seconds. A good moment has one clear idea, starts strong without context, and ends on a line that lands. For each one give me: start and end timestamp, the first sentence spoken, why someone would share it, and a 6-word on-screen title. Skip intros, small talk and anything that needs the previous 10 minutes to make sense. Use only timestamps and words that are in the transcript. If a moment falls between two timestamps, give the nearest one plus its first 5 words so I can find it. If there are fewer than 5 good moments, give me fewer." Prompt 2: A post for each clip "For each clip above, write a post for [X / LinkedIn / Instagram]: a first line that makes someone stop scrolling, 2 lines of context, and one question for the comments. Don't repeat the on-screen title. Under 280 characters for X." That's a 2-hour call turned into a week of posts in under an hour. Most recordings die in a folder because nobody has 2 hours to rewatch them. Let the model rewatch them for you. More prompts like this in the God of Prompt library. 7,200+ prompts, free, no signup. Link in the first reply. Would you let AI pick your best moments, or do you only trust your own cut?

Grok Bot made this Starship Flight 15 highlight reel for me. 🚀 It downloaded the 23-hour livestream, picked out the best moments, edited the clips together, added effects and music, all so quickly. You don’t need a video editor anymore. Just ask Grok Bot.

1:00
God of Prompt avatar
God of Prompt@godofprompt·Nov 15

Practical move: put Decisions in front as a router, Hydrogen in the loop, Carbon only on the hard pass. Burn the Carbon intro price before Dec 1. Don't blindly leave 3.8 Flash until you've measured cost per task. Flash is still cheaper per token through Dec 31.

God of Prompt avatar
God of Prompt@godofprompt·Nov 15

The sleeper is the Decisions API. gemini-decision-1 scores yes/no, choice, and rubric questions in one pass, $0.10 per 1M input, $0 output. OpenAI already ships this idea. Google adds audio, video, PDFs, and 1,000 questions. No caching yet.