← Back to Blog
Free AI Tools I Still Use After Months — Plus the Zero-Cost Student Stack

Free AI Tools I Still Use After Months — Plus the Zero-Cost Student Stack

Most "free AI tools" lists are 72-hour trials and watermarked output. These are the ones I still open after months, plus the student setup — GitHub Student Pack, NotebookLM, Cursor — that covers studying and coding for ₹0.

01A note on "free" AI tools lists

Most lists I've seen include tools that are free for 72 hours, or free with a watermark that makes the output unusable, or technically free but capped at 5 uses per month. I'm not including those.

Everything below is either fully free with no meaningful limit, or has a free tier that's usable for normal work. I use these regularly — I'm a developer based in India and I've been testing AI tools since early 2023, mostly because I couldn't justify paying $20/month for every new thing that launched. Finding free tools that actually work became a habit.

I'm also leaving out tools that look impressive in demos but don't survive normal use. A free tool is only useful if you can come back tomorrow and still get something done. I've been burned by this before — signed up, got excited, hit the paywall after day three. Not including those.

02Writing and thinking: Claude and Gemini

Claude (claude.ai) free tier gives you Claude Sonnet with daily message limits. For most people who aren't doing heavy AI sessions all day, the limit is enough. It's good for writing, coding, research, explaining complex things. I wrote about which Claude prompts actually work for coding if you want to get more out of it.

Gemini (gemini.google.com) is free with a Google account and no message caps on the base model. It's also now embedded in Google Docs and Gmail which is genuinely convenient if you're already working there. Not as strong as Claude for complex reasoning but completely unlimited for free.

Perplexity.ai is free and excellent if you want AI answers with actual cited sources. I use it more like a better search engine than a chatbot.

03Images: three options that are actually free

Most AI image tools are either paid or so limited the free tier is useless. These are the exceptions:

  • Adobe Firefly — free monthly credits, good quality, and the outputs are commercially safe because Adobe trained it on licensed content.
  • Ideogram.ai — free with daily limits. Unusually good at rendering text inside images, which most AI image tools are still terrible at. For banners or anything with readable text, use this one.
  • Microsoft Designer (Bing Image Creator) — free, essentially unlimited for basic use, runs on DALL-E. Not the most flexible but zero cost and decent quality for simple graphics.

04Coding: Cursor and GitHub Copilot

Cursor is a VS Code fork with AI built in. The free tier includes a decent amount of AI completions and chat. If you code regularly it will change your workflow.

GitHub Copilot is free for students and open source maintainers. If you qualify, it's full Copilot access at no cost. Check the GitHub Student Developer Pack — it includes Copilot plus about 20 other tools.

For beginners, the best use is explanation, not autopilot. Ask why an error happened. Ask what a line does. Ask for smaller examples. If you only accept generated code, you'll move faster for a week and get stuck harder later.

05The setup I would choose from zero

If I had to start again with no paid subscriptions, I would use Gemini for unlimited general work, Claude when I need stronger reasoning, Perplexity for research, NotebookLM for documents, Cursor for coding, CapCut for video, and Firefly or Ideogram for images.

That covers writing, studying, coding, research, visuals, and content creation without spending money. More tools than that usually creates clutter. The goal is not to collect AI apps. The goal is to build a workflow you actually remember to use.

For context: I'm working in India on a developer's budget. Paying ₹1500-2000/month for every AI tool that launches is not realistic. This list has to work free, not just technically free. All of these pass that test for me.

06My quick test before keeping a tool

I give every tool the same basic test: can I use it twice in the same week without hitting a paywall, waiting forever, or fighting the interface? If the answer is no, I don't keep it in my workflow.

For writing tools, I test whether the output needs light editing or a full rewrite. For image tools, I check whether the result is usable without obvious artifacts. For coding tools, I check whether the explanation helps me understand the bug, not just paste a fix.

This filter removes most shiny AI products quickly. A tool does not need to be perfect. It needs to be reliable enough that I remember to use it when work is due.

07Three free workflows I use

Blog research workflow: Perplexity for sources, NotebookLM for long PDFs or docs, then Claude or Gemini to pressure-test the outline before I write. I don't ask AI to write the final article from scratch because that's where the content starts sounding flat.

Coding workflow: Cursor for small in-editor changes, Claude free for explaining errors, and Perplexity when I need the latest docs or a current package detail. The prompt I use most is: "explain the cause first, then suggest the smallest fix."

Creator workflow: CapCut for captions and cutting pauses, Canva or Firefly for thumbnails, and ChatGPT for title variations. I usually ask for 20 title ideas, throw away 18, and rewrite the best two in my own voice.

08How to choose without wasting a weekend

Pick one task you do every week and choose the tool for that task. If you write, test Claude and Gemini. If you research, test Perplexity. If you edit video, test CapCut. Don't open twelve accounts in one night.

Use the tool for real work, not just demos. A tool that looks amazing for five minutes may still be annoying after three days. The useful ones become boringly reliable.

After a week, keep the tools you reached for naturally and ignore the rest. That filter is better than any ranking list, including this one.

09Video editing: CapCut

CapCut has become surprisingly capable and most of the AI features are free. Auto captions, background removal, noise reduction, basic AI video generation. If you make short-form content, this is probably the best free option right now.

Runway has a free tier but it's a one-time credit allocation — good for testing, not for ongoing use.

10A few more worth knowing

ElevenLabs — free tier for AI voice generation, decent limits for occasional use. NotebookLM from Google is free and weirdly good for understanding long documents and PDFs. Gamma.app makes presentations from text — free with watermark but saves real time.

Jan.ai is worth knowing if you ever need AI that works completely offline or don't want to send content to a cloud. Free, open-source, runs locally on your machine. Not as smart as Claude but genuinely useful for private work.

For writing use Claude or Gemini. For images use Firefly or Ideogram. For coding use Cursor. For video use CapCut. That covers most of what people actually use AI tools for, and all of it's free.

11If you're a student, check what you already have first

When I was studying, ₹1500/month for any software felt like a lot — not because the tools weren't useful, but because that money came out of food or travel. I ran trial accounts across four email addresses to keep using one tool I needed. Not proud of it, but that was the budget.

The thing most students skip: your college may already pay for this. A lot of universities hold institutional licences for Microsoft Copilot, Grammarly, and Adobe Creative Cloud. Ask the IT department or the library before signing up for anything.

The one not to miss is the GitHub Student Developer Pack. Sign up with your college email at education.github.com and you get GitHub Copilot free plus roughly twenty other tools, including domain names and cloud credits. It is comfortably the best student deal in tech right now, and most students never claim it.

12The student stack that costs nothing

For studying: NotebookLM with your lecture notes, textbook chapters, and past papers. Because it answers only from what you upload, the output stays inside your actual syllabus instead of drifting into generic internet material. Ask it to quiz you, answer without looking, then ask where your answer was weak.

For research: Perplexity instead of Google, so you get citations you can actually verify and cite in your own work.

For coding assignments: Cursor on the free tier. When you're stuck on an error at midnight before a submission, having it explain what's wrong is genuinely useful — but ask for the cause and a hint, not the full solution. It feels slower and it builds the muscle you need when there's no AI in the exam hall.

For slides and design: Gamma.app generates presentations from text (free tier has a watermark), and Canva's free tier covers background removal and basic image generation.

13What I'd tell a student to avoid

Anything promising to write full assignments. It creates two problems at once: academic risk now, and a gap in understanding that shows up later. I've seen people get caught in vivas because they couldn't explain their own submitted work.

Paying for five subscriptions because five different YouTubers each called a different tool essential. Start free, use them for a month, then pay only for the one that's clearly saving you time every week.

Collecting tools instead of using them. The best student setup is boring: one research tool, one study-from-your-own-notes tool, one writing assistant, one coding helper. More than that is procrastination dressed as productivity.

14Frequently Asked Questions

Abhinav Sinha

Written by

Abhinav Sinha

Software engineer at Varahe Analytics, where I build a multi-LLM platform running in production across nine Indian languages. I also run @9_face_toon, a Hindi animation channel with 40K+ followers. I write about problems from both, not about product launches.

Testing, sourcing, disclosures, and corrections