Which AI should you use today?
Eight tools. One honest comparison. Best use-cases, real pricing, and a copy-paste starter prompt for each — so you stop paying for five subscriptions you don't use.
"The Swiss Army knife of AI."
"The thoughtful writer that actually reads your doc."
"AI that lives inside your Google life."
"The AI that's been doomscrolling X."
"The art director that never sleeps."
"Type a feeling, get a song."
"The voice that sounds like a person."
"Prompt it. Ship it."
"Sora quality. TikTok speed. Coffee-money price."
"Google, if Google actually answered."
"VS Code, but it reads your whole repo."
"Cinema, from one paragraph."
"The video editor that also generates."
"Photoreal, and it can spell."
"Your AI news anchor, in 175 languages."
"GPT-class reasoning at open-source prices."
"The video model that started the arms race."
"The physics champion of AI video."
"Your sources. Your notebook. Real citations."
"Finally, an image AI that can spell."
"shadcn UI, from a sentence."
"The agent that finishes the whole task."
"Real-time canvas for AI creators."
"Cinematic camera moves, one click."
"Brand-safe vectors and mockups."
The 30-second comparison
| Tool | Best at | Free tier | Paid from | Category |
|---|---|---|---|---|
ChatGPT OpenAI | Daily writing & email | $0 | $20/mo | Chat |
Claude Anthropic | Long documents (PDFs, contracts) | $0 | $20/mo | Chat |
Gemini Google | Anything with Google Workspace | $0 | $20/mo | Chat |
Grok xAI | Live news & trends | $0 | $30/mo | Chat |
Midjourney Midjourney Inc. | Posters & marketing visuals | $10/mo | $30/mo | Image |
Suno Suno AI | Brand jingles & ads | $0 | $10/mo | Music |
ElevenLabs ElevenLabs | YouTube/podcast voiceover | $0 | $5/mo | Voice |
Lovable Lovable | MVPs and landing pages in hours | $0 | $25/mo | Builder |
Seedance 2.0 ByteDance | High-volume TikTok / Reels / Shorts | Free | $9.99/mo | Video |
Perplexity Perplexity AI | Cited research | $0 | $20/mo | Chat |
Cursor Anysphere | Full-repo refactors | $0 | $20/mo | Builder |
Veo 3 Google DeepMind | Cinematic hero shots | $20/mo | $250/mo | Video |
Runway Runway | Motion-brush precise animation | $0 | $15/mo | Video |
Flux Black Forest Labs | Photoreal product shots | Free | ~$0.04/img | Image |
HeyGen HeyGen | Training videos & LMS content | $0 | $29/mo | Video |
DeepSeek DeepSeek | Cheap high-volume API calls | Free | $0.27/M in · $1.10/M out | Chat |
Sora OpenAI | Story-driven short films | $20/mo | $200/mo | Video |
Kling Kuaishou | Realistic human motion | $0 | $10/mo | Video |
NotebookLM Google | Studying long PDFs & courses | $0 | $20/mo | Chat |
Ideogram Ideogram | Posters with headlines | $0 | $8/mo | Image |
v0 Vercel | Landing pages & marketing sites | $0 | $20/mo | Builder |
Manus Butterfly Effect | Long multi-step research | $0 | $19/mo | Chat |
Krea Krea AI | Real-time sketch-to-image | $0 | $10/mo | Image |
Higgsfield Higgsfield AI | Cinematic camera moves | $0 | $9/mo | Video |
Recraft Recraft | SVG / vector output | $0 | $12/mo | Image |
ChatGPT
"The Swiss Army knife of AI."
The most popular all-rounder. Great at reasoning, writing, coding, and image generation in one chat.
If in doubt — ask ChatGPT first. It's the friend who knows a little about everything.
Best for
- Daily writing & email
- Brainstorming ideas
- Coding helper
- Summaries & translation
- Built-in image generation
Not for
- —Latest niche research (use Perplexity)
- —Long agentic coding (use Claude)
Pricing
| Free | $0 | GPT-5 mini, daily limits |
| Plus | $20/mo | GPT-5, image gen, voice |
| Pro | $200/mo | Unlimited GPT-5 Pro, deeper reasoning |
NAX Score
You are my senior advisor. Before answering, ask me 3 clarifying questions about: my goal, my audience, and my constraints. Then give me a step-by-step plan and one example.
Pro tips
- →Use Projects to keep context between chats.
- →Say 'think step by step' to unlock reasoning.
- →Upload a screenshot — it can read UI and suggest fixes.
Claude
"The thoughtful writer that actually reads your doc."
Best long-context model. Reads a 200-page PDF without panicking. Coders' favorite for big refactors.
Claude = the calm senior who reads everything before talking.
Best for
- Long documents (PDFs, contracts)
- Production-quality writing
- Big codebase refactors
- Nuanced reasoning
Not for
- —Image generation (it can't)
- —Latest news (no live web by default)
Pricing
| Free | $0 | Sonnet, light usage |
| Pro | $20/mo | Sonnet 4.5 + Opus, projects |
| Max | $100–200/mo | 5–20x usage, priority |
NAX Score
I'm attaching a long document. Do three passes: (1) one-paragraph TL;DR, (2) the 5 most important findings with quotes, (3) 3 risks I might miss. Be ruthlessly concise.
Pro tips
- →Drop entire codebases as a .zip — it handles 200K tokens.
- →Ask it to 'be skeptical' — it pushes back well.
- →Use Artifacts to get a live preview of code/HTML.
Gemini
"AI that lives inside your Google life."
Massive 2M-token context, plugged into Gmail, Docs, YouTube, and Search. Free tier is unusually generous.
Gemini = the Google intern that already has your calendar open.
Best for
- Anything with Google Workspace
- Watching/summarizing YouTube
- Huge context windows
- Free fast model
Not for
- —Edgy/creative writing (it plays safe)
- —Heavy coding (use Claude)
Pricing
| Free | $0 | Gemini Flash, generous limits |
| AI Pro | $20/mo | Gemini 3 Pro, Deep Research |
| AI Ultra | $250/mo | Veo video, 30TB storage |
NAX Score
Run Deep Research on [topic]. Compare top 5 sources, flag contradictions, then write me a 1-page brief with citations and 3 'what most people miss' insights.
Pro tips
- →Paste a YouTube link → ask for a summary with timestamps.
- →@Gmail and @Drive let it search your own files.
- →Use Deep Research for any topic you'd normally Google for an hour.
Grok
"The AI that's been doomscrolling X."
Real-time access to X (Twitter) and the open web. Cheekier tone, fewer filters, image gen built in.
Grok = the friend who saw the meme before you did.
Best for
- Live news & trends
- Social listening on X
- Edgy/comedy writing
- Quick image gen
Not for
- —Enterprise serious work
- —Long-context analysis
Pricing
| Free | $0 | Limited Grok 4 access |
| SuperGrok | $30/mo | More usage, Voice |
| Heavy | $300/mo | Multi-agent Grok 4 Heavy |
NAX Score
Scan X right now for [topic]. Give me: top 3 trending takes, the loudest dissent, one underrated post, and what a smart marketer would do with this in 24 hours.
Pro tips
- →Ask 'what is X saying about ___ right now' for live pulse.
- →Use the Fun/Unhinged mode for marketing hooks.
- →Bundled free with X Premium — check before paying twice.
Midjourney
"The art director that never sleeps."
Still the most cinematic, painterly image model. The default for posters, moodboards, and album covers.
Midjourney = the artist. ChatGPT image = the intern. Pick the artist when it matters.
Best for
- Posters & marketing visuals
- Concept art & moodboards
- Brand-quality portraits
- Stylized illustrations
Not for
- —Accurate text inside images
- —Photoreal product shots (try Flux)
Pricing
| Basic | $10/mo | ~200 images |
| Standard | $30/mo | Unlimited relaxed |
| Pro | $60/mo | Stealth mode, more fast hours |
NAX Score
[subject], [action], [environment], shot on Arri Alexa 65mm, soft rim light, shallow depth of field, cinematic color grade, hyperreal, --ar 16:9 --style raw --v 7
Pro tips
- →Always end with --ar (aspect ratio) and --style raw for photoreal.
- →Reference a real photographer's name for instant style.
- →Use /describe on an image to reverse-engineer its prompt.
Suno
"Type a feeling, get a song."
Generates full songs — vocals, lyrics, production — from a one-line prompt. Wild for jingles and demos.
Suno = karaoke that writes itself. Bad idea + Suno = unexpectedly great idea.
Best for
- Brand jingles & ads
- Demo songs & sketches
- Background music for videos
- Personalized birthday songs
Not for
- —Pro-mastered release tracks
- —Exact instrument control
Pricing
| Free | $0 | 10 songs/day, non-commercial |
| Pro | $10/mo | 500 songs, commercial use |
| Premier | $30/mo | 2,000 songs, priority |
NAX Score
Style: upbeat indie pop, female vocal, 90 BPM, claps + warm synth. Lyrics: a 30-second jingle about [brand] that ends with the tagline '[tagline]'. Catchy hook, 2 verses, 1 chorus.
Pro tips
- →Use [Verse] [Chorus] [Bridge] tags inside lyrics for structure.
- →Describe production, not genres — 'warm 70s vinyl' beats 'pop'.
- →Generate 4 variants and keep the best 8 seconds, not the best song.
ElevenLabs
"The voice that sounds like a person."
The gold standard for AI voice — narration, dubbing, voice cloning, and conversational agents.
ElevenLabs = your voice actor on retainer, in 32 languages, awake at 3am.
Best for
- YouTube/podcast voiceover
- Audiobooks
- Dubbing into other languages
- Customer-facing voice agents
Not for
- —Singing (use Suno)
- —Real-time low-latency calls under 200ms (it's close)
Pricing
| Free | $0 | 10 min/mo, non-commercial |
| Starter | $5/mo | 30 min, voice cloning |
| Creator | $22/mo | 100 min, pro voices |
NAX Score
Voice: warm, mid-30s, conversational. Pace: medium with natural pauses. Use ellipses... for breath. Use ALL CAPS sparingly for emphasis. Read this like you're explaining it to a smart friend over coffee: [your script]
Pro tips
- →Lower the Stability slider for more emotion, raise for consistency.
- →Add commas and ellipses to control pacing — punctuation IS direction.
- →Clone your own voice once — reuse for every video forever.
Lovable
"Prompt it. Ship it."
The AI app builder that turns plain-English prompts into real production web apps — frontend, database, auth, payments, deploy.
Lovable = your full-stack team in a chat box. NAX is an official Lovable Solution Partner.
Best for
- MVPs and landing pages in hours
- Internal tools & dashboards
- Full-stack apps with auth + DB
- Founders who can't code (yet)
- Agencies shipping client work 5× faster
Not for
- —Native mobile (iOS/Android shells)
- —Heavy 3D / game engines
- —Pixel-perfect Figma handoff for a 50-person design team
Pricing
| Free | $0 | Daily messages, public projects |
| Pro | $25/mo | Private projects, custom domains, more credits |
| Teams | $30/seat | Shared workspace, roles, SSO |
NAX Score
Build a [type of app] for [audience] that lets users [main action]. Context: - We are [company] and our voice is [tone]. - Brand colors: [#hex, #hex]. Font feel: [modern serif / clean sans / etc]. Scope (v1 only): - Pages: [Home, Pricing, Dashboard]. - Auth: email + Google. - Database: tables for [users, projects, items] with sensible fields. Design: - Apple-clean, generous spacing, real product imagery, no emoji. Acceptance: - I can sign up, create a [item], see it in my dashboard, and log out. - Mobile looks as good as desktop. Do NOT add features outside this scope. Ask me 3 clarifying questions first if anything is unclear.
Pro tips
- →Always write Context → Goal → Scope → Design → Acceptance. Vague prompts = vague apps.
- →Use Lovable Cloud for auth, DB and storage — no Supabase setup needed.
- →When stuck, paste the error verbatim and ask: 'What's the smallest change to fix this?'
- →Connect GitHub and a custom domain once — then ship updates in minutes.
- →NAX runs the official Lovable masterclass (Chapter 13) — prompt anatomy, recipes, debugging, ejection.
Seedance 2.0
"Sora quality. TikTok speed. Coffee-money price."
ByteDance's flagship AI video model. 1080p in seconds, native multi-shot in one prompt, and roughly 5–10× cheaper than Sora or Veo.
The company that made you scroll for hours now makes the fastest way to fill that scroll — with your own videos.
Best for
- High-volume TikTok / Reels / Shorts
- Multi-shot sequences in one generation
- Fast product & ad b-roll on a budget
- Image-to-video with consistent characters
Not for
- —Long narrative story films (use Sora)
- —Legible on-screen text (add in CapCut)
- —Regulated / brand-sensitive industries
Pricing
| Doubao / Jimeng | Free | ~5 free videos/day in-app |
| CapCut Pro | $9.99/mo | Bundled with CapCut features |
| API (Volcano/Replicate) | ~$0.03/s | 720p; ~$0.06/s for 1080p |
NAX Score
Shot 1 (2s, slow dolly in): [subject + action]. [Environment]. [Lighting]. [Mood]. Shot 2 (2s, macro): [subject + action]. [Environment]. [Lighting]. [Mood]. Shot 3 (2s, medium): [subject + action]. [Environment]. [Lighting]. [Mood]. Overall style: Kodak Portra 400, shallow depth of field, cinematic warm grade. Aspect: 9:16. Total 6s.
Pro tips
- →Iterate at 720p, only re-render the winner at 1080p.
- →Feed a reference image — character consistency across shots is unusually good.
- →Write shot durations that add up to your target length — Seedance respects them.
- →If motion feels sped-up, add 'real-time pacing, no time-lapse'.
Perplexity
"Google, if Google actually answered."
AI answer engine with real citations. Best-in-class for research, comparison shopping, and 'what's the truth' questions.
Perplexity = the librarian. Cites every source, never bluffs.
Best for
- Cited research
- Real-time news
- Product comparison
- Academic-style questions
Not for
- —Long-form creative writing
- —Coding help (use Claude/ChatGPT)
Pricing
| Free | $0 | Standard search, limited Pro |
| Pro | $20/mo | Pro Search, GPT-5, Claude, file uploads |
| Enterprise | $40/seat | SSO, admin, spaces |
NAX Score
Research [topic]. Give me: (1) the current consensus with 3 sources, (2) one strong dissenting view with source, (3) a 5-bullet summary I could quote in a meeting. Cite everything inline.
Pro tips
- →Use 'Focus' → Academic for research, Reddit for real-user opinion.
- →Pro Search does multi-step queries — worth the wait.
- →Ask 'compare X vs Y with citations' — it beats plain Google every time.
Cursor
"VS Code, but it reads your whole repo."
AI-first code editor. Understands your entire codebase, refactors in agent mode, and writes tests you'd actually keep.
Cursor = pair programmer that already read the docs and your git history.
Best for
- Full-repo refactors
- Agent-mode multi-file edits
- TypeScript / React / Python codebases
- Onboarding to a new codebase
Not for
- —No-code beginners (use Lovable)
- —Non-code writing
Pricing
| Hobby | $0 | Limited fast requests |
| Pro | $20/mo | 500 fast + unlimited slow |
| Business | $40/seat | Team billing, SSO, privacy mode |
NAX Score
In agent mode: audit [feature] across the repo. Propose the smallest safe refactor. Show a diff plan first, then apply — one file at a time, running the test suite after each. Stop and ask if anything is ambiguous.
Pro tips
- →Cmd+K for inline edits, Cmd+L for chat, Cmd+I for agent.
- →@codebase to search whole repo, @docs to add framework docs.
- →Turn on Privacy Mode if working on client code — nothing is retained.
Veo 3
"Cinema, from one paragraph."
Google's flagship video model. 4K, native audio (dialogue + SFX + music), unmatched physics realism and character consistency.
Veo = the film school graduate. Slow, expensive, but Oscar-tier.
Best for
- Cinematic hero shots
- Product films with dialogue
- Brand storytelling
- Photoreal character scenes
Not for
- —Fast iteration on a budget (use Seedance)
- —Text inside frame
Pricing
| Gemini AI Pro | $20/mo | Veo 3 Fast, limited daily gens |
| AI Ultra | $250/mo | Veo 3 Quality, higher limits, 4K |
| Vertex API | $0.35–0.75/s | Enterprise, commercial license |
NAX Score
A [subject] [action] in [environment]. Camera: [slow dolly / handheld / crane]. Lighting: [golden hour / neon / soft key]. Lens: 35mm anamorphic, shallow DOF. Style: Kodak Vision3 500T, cinematic grade. Audio: subtle ambient [sound], one line of dialogue: '[line]'. 8 seconds.
Pro tips
- →Write audio direction in the same prompt — Veo 3 generates SFX/dialogue.
- →Feed a reference image for character consistency across scenes.
- →Start with Veo Fast to iterate, then upscale the keeper in Quality.
Runway
"The video editor that also generates."
Gen-4 video model plus a full pro editing suite — motion brush, camera control, green-screen, lip-sync, all in one browser tab.
Runway = Adobe for the AI era. Not just a generator — an editor.
Best for
- Motion-brush precise animation
- AI VFX (rotoscope, inpaint)
- Lip-sync to any audio
- Camera-controlled shots
Not for
- —Free-tier heavy use (credits burn fast)
- —Native audio (add in post)
Pricing
| Free | $0 | 125 credits, watermarked |
| Standard | $15/mo | 625 credits, no watermark |
| Pro | $35/mo | 2,250 credits, 4K, private |
NAX Score
Base image: [describe]. Use motion brush to move [subject] slowly [direction]. Camera: 3D orbit right at slow speed. Duration: 5s. Style: filmic, subtle grain, real-time pacing.
Pro tips
- →Motion Brush > text prompts for precise movement.
- →Use Frames as your image-to-video keyframes for consistency.
- →Act-One clones any actor's expressions onto your character.
Flux
"Photoreal, and it can spell."
Open-weight image model that beats Midjourney at photorealism and legible in-image text. The pros' choice for product shots.
Flux = the photographer. Midjourney = the painter. Pick per job.
Best for
- Photoreal product shots
- Legible text inside images
- Editorial portraits
- Brand-consistent characters
Not for
- —Painterly / stylized art (use Midjourney)
- —Free unlimited (self-host or pay per gen)
Pricing
| Schnell (open) | Free | Self-host / HuggingFace / Replicate |
| Pro (API) | ~$0.04/img | Via Replicate, fal, Together |
| Ultra + Kontext | ~$0.06/img | Best fidelity, image-to-image edits |
NAX Score
Studio photograph of [product] on [surface], [lighting: soft North-facing window / three-point studio]. Shot on Phase One IQ4 150MP, 80mm f/2.8, natural color, shadow detail preserved. Include legible text on packaging: '[exact text]'.
Pro tips
- →Flux Kontext is unmatched at editing existing photos with a prompt.
- →Use exact camera + lens language — Flux respects it more than MJ.
- →For type inside images, quote the exact text — it renders it correctly ~90% of the time.
HeyGen
"Your AI news anchor, in 175 languages."
Talking-avatar and video-translation platform. Turn a script into a lip-synced presenter video — or dub an existing video into 30+ languages with matched mouth movement.
HeyGen = the presenter that never asks for a raise.
Best for
- Training videos & LMS content
- Multilingual marketing localisation
- Avatar-led product demos
- Personal-avatar YouTube channels
Not for
- —Cinematic B-roll (use Veo/Seedance)
- —Complex acting/emotion (limited range)
Pricing
| Free | $0 | 3 videos/mo, watermarked |
| Creator | $29/mo | Unlimited HD, no watermark |
| Team | $89/seat | Custom brand avatars, 175 lang |
NAX Score
Avatar: [pick a HeyGen preset or your cloned self]. Voice: warm, conversational, 150 wpm. Script (English): [90-second explainer of your product's #1 benefit + one CTA]. Generate + translate the same avatar into: Malay, Chinese (Mandarin), Bahasa Indonesia, Thai.
Pro tips
- →Clone your own avatar once — 2-minute selfie video is enough.
- →Use the Interactive Avatar API for real-time customer support agents.
- →Video Translate keeps voice + lip-sync — killer for global launches.
DeepSeek
"GPT-class reasoning at open-source prices."
Chinese open-weight model that matches frontier labs on math and code — free on the web, and ~10–20× cheaper than GPT on API.
DeepSeek = the honors student who charges pocket money.
Best for
- Cheap high-volume API calls
- Math & competition-level reasoning
- Chinese + English bilingual work
- Self-hosting the open weights
Not for
- —Politically sensitive topics (soft-filtered)
- —Image generation
Pricing
| Web Chat | Free | V3 + R1 reasoning, unlimited |
| API — chat | $0.27/M in · $1.10/M out | Cache hits ~10× cheaper |
| API — reasoner | $0.55/M in · $2.19/M out | R1 chain-of-thought |
NAX Score
Solve this step by step, then verify by working the answer backward. If the two paths disagree, flag it and try again. Problem: [paste problem]
Pro tips
- →Toggle 'DeepThink (R1)' for math/logic; leave off for speed.
- →The weights are open — you can run V3 locally on a big GPU.
- →Great translator between English ↔ Simplified Chinese for docs.
Sora
"The video model that started the arms race."
OpenAI's flagship video model. Best-in-class physics, narrative continuity, and remix workflow inside ChatGPT.
Sora = the director's cut. Fewer takes, better takes.
Best for
- Story-driven short films
- Character-consistent scenes
- Remix / storyboard workflows
- 1080p vertical for social
Not for
- —High-volume cheap b-roll (use Seedance)
- —Real-time generation
Pricing
| ChatGPT Plus | $20/mo | Sora, 480p/720p, limited length |
| ChatGPT Pro | $200/mo | Sora Pro, 1080p, 20s, higher limits |
NAX Score
A [character description] [core action] in [location]. Beat 1 (0–3s): [establishing detail]. Beat 2 (3–7s): [emotional shift]. Beat 3 (7–10s): [payoff]. Camera: [movement]. Lens: 35mm, shallow DOF. Style: Kodak Vision3, cinematic grade.
Pro tips
- →Use the storyboard tool — write beats, not a wall of adjectives.
- →Feed a still frame to lock character look across generations.
- →Remix a good clip 3–4 times before starting from scratch.
Kling
"The physics champion of AI video."
Kuaishou's video model. Best-in-class motion realism, lip-sync, and multi-character scenes — a fan favorite for dance, fights, and food shots.
Kling = the stunt coordinator. If it moves, Kling moves it best.
Best for
- Realistic human motion
- Action / dance / sports shots
- Lip-synced dialogue clips
- Image-to-video with elements
Not for
- —Long narrative films (max ~2 min)
- —Legible on-screen text
Pricing
| Free | $0 | 166 daily credits, watermarked |
| Standard | $10/mo | 660 credits, no watermark |
| Pro | $37/mo | 3,000 credits, 1080p, priority |
NAX Score
[Subject] performs [specific physical action] in [environment]. Camera tracks alongside at medium distance. Real-time pacing, no slow-motion. Natural physics: weight, momentum, cloth movement. 5 seconds. 9:16.
Pro tips
- →Use 'Elements' to pin a face + prop + background across clips.
- →Kling 2.1 Master mode is worth the extra credits for hero shots.
- →Say 'real-time pacing' — default output tends to feel slow-mo.
NotebookLM
"Your sources. Your notebook. Real citations."
Upload PDFs, docs, YouTube, and URLs — get a grounded research assistant that only answers from your sources, plus an AI-hosted podcast of them.
NotebookLM = the study group that actually did the reading.
Best for
- Studying long PDFs & courses
- Turning research into a podcast
- Meeting-notes → briefs
- Grounded citations only
Not for
- —Open-web research (use Perplexity)
- —Creative writing without sources
Pricing
| Free | $0 | 100 notebooks, 50 sources each |
| AI Pro | $20/mo | 500 notebooks, 5× audio, video overviews |
NAX Score
From my sources only: (1) 5-bullet TL;DR, (2) 3 counter-arguments the authors miss, (3) the 10 quotes I'd cite in a talk — each with source + page. Do not add outside knowledge.
Pro tips
- →Generate the Audio Overview — 2 AI hosts turn any topic into a 15-min podcast.
- →Add YouTube URLs as sources — it reads transcripts.
- →Use it as an exam-prep tutor: 'quiz me on Sources 2 and 4'.
Ideogram
"Finally, an image AI that can spell."
The image model to beat for typography — logos, posters, and packaging with legible text on the first try.
Ideogram = the graphic designer. Midjourney paints, Ideogram letters.
Best for
- Posters with headlines
- Logo concepts & wordmarks
- Packaging mockups
- Social ads with copy
Not for
- —Cinematic photoreal (use Flux)
- —Painterly art (use Midjourney)
Pricing
| Free | $0 | 10 slow gens/day |
| Basic | $8/mo | 400 fast gens, private |
| Plus | $20/mo | 1,000 fast, upscaling, no watermark |
NAX Score
Bold minimalist poster. Headline text: '[EXACT COPY]' in a chunky grotesque sans, 3-color palette [#hex, #hex, #hex]. Subtle geometric background. Print-ready, high contrast, 2:3.
Pro tips
- →Quote the exact text in single quotes — Ideogram respects it.
- →Use the 'Magic Prompt' toggle off when you want your wording preserved.
- →Great for A/B ad variations — remix keeps text, changes visuals.
v0
"shadcn UI, from a sentence."
Vercel's generative UI tool. Prompt to production-quality Next.js + Tailwind + shadcn components, deploy in one click.
v0 = the front-end intern who ships pixel-perfect the first time.
Best for
- Landing pages & marketing sites
- Dashboards with shadcn
- Design → code handoff
- One-click Vercel deploy
Not for
- —Full-stack DB apps (use Lovable)
- —Non-React stacks
Pricing
| Free | $0 | Daily message limit |
| Premium | $20/mo | 5,000 messages, private |
| Team | $30/seat | Shared workspace, roles |
NAX Score
Build a [section: pricing / hero / dashboard] using shadcn, Tailwind, and lucide icons. Style: Apple-clean, generous whitespace, dark mode default. Include: [3 tiers with feature list] / [KPI cards + chart]. Fully responsive. No lorem ipsum — use real, plausible copy.
Pro tips
- →Ask for 3 variants — v0 nails one of them almost every time.
- →Paste a screenshot of a site you like — it reverse-engineers layout.
- →Export the exact shadcn components straight into your repo.
Manus
"The agent that finishes the whole task."
General-purpose autonomous agent. Give it a goal — book flights, build a report, scrape and analyze — and it browses, codes, and delivers files.
Manus = the intern that never asks 'what next?' — it just does it.
Best for
- Long multi-step research
- Scraping + spreadsheet building
- Trip / event planning
- Deliverable-heavy tasks
Not for
- —Real-time chat
- —Sensitive data on strict compliance
Pricing
| Free | $0 | Daily task credits |
| Starter | $19/mo | More parallel tasks + fast lane |
| Pro | $39/mo | Highest limits, priority browsers |
NAX Score
Goal: [outcome]. Deliverables: (1) a CSV of [rows], (2) a 1-page summary, (3) a slide deck. Sources: browse [sites]. Constraints: [budget, dates]. Stop and confirm before spending credits on step 3.
Pro tips
- →Always name the deliverable file types — 'return as .xlsx + .pdf'.
- →Break huge goals into sub-tasks — cheaper and more accurate.
- →Check its browser recordings — you'll learn how it thinks.
Krea
"Real-time canvas for AI creators."
Live image + video playground. Sketch, generate, upscale, and animate — all in one visual canvas with real-time preview.
Krea = Figma × Photoshop × Midjourney in one tab.
Best for
- Real-time sketch-to-image
- Upscaling & enhance
- Mixing multiple models in one flow
- Motion + image in one canvas
Not for
- —Batch API workloads
- —Offline / self-host
Pricing
| Free | $0 | Daily gens, watermarked |
| Basic | $10/mo | No watermark, priority |
| Pro | $35/mo | All models, higher res, video |
NAX Score
Start with a rough sketch of [scene]. Style transfer to [photoreal Kodak Portra 400 / anime cel-shade / brutalist collage]. Keep composition, upgrade lighting and materials. Then upscale 4×.
Pro tips
- →Use Realtime mode — every brush stroke updates the render.
- →Combine Flux + SDXL in the same session for hybrid looks.
- →Enhancer upscaler is one of the best on the market for portraits.
Higgsfield
"Cinematic camera moves, one click."
AI video model built around real cinematography — dolly zooms, crash zooms, bullet time, and 50+ camera controls you'd need a rig for.
Higgsfield = the camera operator you couldn't afford.
Best for
- Cinematic camera moves
- Music video B-roll
- Product hero shots with motion
- Social ads that feel expensive
Not for
- —Dialogue-heavy scenes
- —Long-form storytelling
Pricing
| Free | $0 | Weekly credits, watermarked |
| Basic | $9/mo | 150 credits, no watermark |
| Pro | $29/mo | 600 credits, all camera presets |
NAX Score
Base image: [describe]. Camera preset: 'Bullet Time' → hold on subject 2s, then dolly-zoom out. Real-time pacing. Style: 35mm anamorphic, filmic grain, cinematic grade. 5 seconds.
Pro tips
- →Pick the camera preset FIRST — it defines the shot.
- →Feed a strong reference image; Higgsfield preserves composition well.
- →Combine two clips with matching motion for seamless transitions.
Recraft
"Brand-safe vectors and mockups."
Design-focused AI — vector output, brand style sets, and product mockups a designer would actually ship.
Recraft = the in-house brand designer who never gets tired.
Best for
- SVG / vector output
- Brand style consistency
- Icons, illustrations, mockups
- Editorial infographics
Not for
- —Cinematic photoreal
- —Video
Pricing
| Free | $0 | 50 daily credits |
| Basic | $12/mo | 1,000 credits, private |
| Advanced | $36/mo | 5,000 credits, style sets, API |
NAX Score
Set of 12 flat vector icons in a consistent style: rounded 2px stroke, 2-color palette [#hex, #hex], no gradients, minimal geometric shapes. Concepts: [list 12]. Export SVG.
Pro tips
- →Create a Style once, reuse across 100+ assets for brand consistency.
- →Export true SVG — infinitely scalable for print and web.
- →Great for turning rough sketches into clean vector logos.
Now go write better prompts.
The tool matters less than the prompt. Grab our reusable templates — system, developer, and user — in the Prompt Cookbook.
