30+

TOOLS COVERED

$27.5K

IN FREE CREDITS

20

TOKEN HACKS

$0

STUDENT STACK

95%

MAX SAVINGS

2026

UPDATED APRIL

PART 1

The Full Map

30+ AI coding tools sorted by category

PART 2

The Free Credit Vault

Every program, exact value, how to apply

PART 3

Token-Saving Playbook

20 power-user tricks to cut costs 50-95%

PART 4

The 3 Stacks

Student ($0) / Founder ($40) / Power User ($200+)

PART 5

Do This Today

7-step action plan to claim your credits

THE HEADLINE NUMBER

A broke student with zero dollars can stack free credits worth roughly $27,500 across Anthropic, OpenAI, AWS, GitHub, Cursor, Vercel, and JetBrains. Everything you need to build a startup. For free. The only cost is 30 minutes of applications.

HOW TO USE THIS DOC

• First read: skim Part 1 to know what exists
• Second read: go to Part 2 and apply for every credit you qualify for
• Bookmark: Part 3 — come back every time your bill scares you
• Forward: Part 4 to anyone asking "what tools should I use"

01. Claude Code

Anthropic's CLI + IDE for agentic coding. The gold standard for reasoning.

Pricing: Pro $20/mo · Max $100-200/mo · API pay-as-you-go

Edge: Best reasoning model on the planet. Subagents, MCP, hooks, and prompt caching built in.

Link: claude.com/code

02. Cursor

AI-first fork of VS Code. Multi-model (Claude, GPT, Gemini).

Pricing: Hobby free · Pro $20/mo · Pro+ $40 · Ultra $200

Edge: Best UX in the game. Tab autocomplete is magic. Student plan = 1 year Pro FREE.

Link: cursor.com

03. Windsurf

Codeium's IDE with its own SWE-1.5 model. Stupidly fast agents.

Pricing: Free tier · Pro $20/mo · Max $200/mo

Edge: Cheapest unlimited agent tier. ~13x faster than Sonnet on Codeium's own benchmark.

Link: windsurf.com

04. Google Antigravity

Google's new AI IDE (Nov 2025). Multi-agent orchestration with Gemini 3.1 Pro.

Pricing: FREE (public preview, generous limits)

Edge: Run multiple agents in parallel. Free access to Gemini 3.1 Pro. Public preview = zero cost right now.

Link: antigravity.google

05. GitHub Copilot

The OG. Now multi-model (Claude + GPT + Gemini). Free tier is real.

Pricing: Free (2K completions/mo + 50 chats) · Pro $10/mo

Edge: Cheapest paid tier anywhere. Free for every verified student. Included in GitHub Student Pack.

Link: github.com/features/copilot

06. Zed

Rust-based, hyper-fast editor with built-in AI.

Pricing: Free · AI features optional add-on

Edge: Lowest-latency coding experience. Built-in collaboration.

Link: zed.dev

07. JetBrains AI Assistant

AI inside IntelliJ / PyCharm / WebStorm / etc.

Pricing: Bundled with IDE · Free for students via GitHub Pack

Edge: Only option if you're a JetBrains loyalist. Refactoring-aware.

Link: jetbrains.com/ai

08. Cline (ex-Claude Dev)

Open-source VS Code extension. Agentic. BYO API key.

Pricing: Free (you pay for your own API calls)

Edge: Most transparent agent in the market. Auto-approves safe actions.

Link: github.com/cline/cline

09. Continue.dev

Open-source VS Code + JetBrains extension. Run any model, local or API.

Pricing: 100% FREE · BYO keys (Ollama, Groq, OpenRouter, etc.)

Edge: The #1 free option. Plug in free Groq / local Ollama and pay nothing.

Link: continue.dev

10. Aider

Terminal AI pair programmer. Git-aware. Supports Claude, GPT, Gemini.

Pricing: Free (BYO API) · Prompt caching built-in

Edge: --cache-prompts gives you 90% savings. Best terminal workflow.

Link: aider.chat

11. Cody (Sourcegraph)

Codebase-aware AI with huge context windows.

Pricing: Pro $9/mo · Free tier status = verify (changed June 2025)

Edge: 1M token context on Pro. Best for massive monorepos.

Link: sourcegraph.com/cody

12. Supermaven

Autocomplete only. No chat. Just stupidly fast completion.

Pricing: FREE forever (unlimited completions) · Pro $10/mo

Edge: The only serious autocomplete with a forever-unlimited free tier. 1M context.

Link: supermaven.com

13. Tabnine

Privacy-first autocomplete. Trains on your private code.

Pricing: Dev $12/mo · Enterprise $39+ · Free tier status = verify

Edge: Best for teams that can't send code to third parties.

Link: tabnine.com

14. Gemini Code Assist

Google's IDE plugin powered by Gemini.

Pricing: Free tier with rate limits · Paid tiers available

Edge: Free tier is genuinely usable. Tight integration with Google Cloud.

Link: codeassist.google

15. Amazon Q Developer

AWS's coding AI (formerly CodeWhisperer). IDE + CLI.

Pricing: Free tier available · Enterprise custom

Edge: Built-in security scanning. Best if you're deep in AWS.

Link: aws.amazon.com/q/developer

16. Warp Terminal

AI-native terminal. Inline command generation and error fixing.

Pricing: Free (75 AI credits/mo, 150 first 2 months) · Build $15/mo

Edge: Changes how you think about terminals. Zero-data-retention with LLM providers.

Link: warp.dev

17. Bolt.new

Full-stack apps from prompts in your browser. Built on StackBlitz.

Pricing: Free (300K tokens/day, 1M/mo) · Pro $20/mo

Edge: Instant preview. Deploy in one click. Best for rapid prototyping.

Link: bolt.new

18. v0 by Vercel

Generate React + Tailwind + shadcn components from text.

Pricing: Free ($5/mo baseline + $2/day login bonus) · Premium custom

Edge: Ships straight into Vercel. Best component generation quality in the market.

Link: v0.dev

19. Lovable

Full app generator. No-code-inclined builders love it.

Pricing: Free (30 credits/mo, 5/day) · Pro $25/mo

Edge: Most polished end-to-end experience. Great for non-devs.

Link: lovable.dev

20. Replit Agent

Cloud IDE + agent that builds multi-file projects for you.

Pricing: Starter free (limited) · Core $20/mo

Edge: Everything lives in the cloud. Nothing to install. Perfect for school Chromebooks.

Link: replit.com

21. Same.new

App builder in browser. Emerging challenger.

Pricing: Verify current pricing at same.new

Edge: Lighter-weight than Bolt. Verify free tier before committing.

Link: same.new

22. Softgen.ai

Business logic generator. Think CRUD apps from specs.

Pricing: Verify current pricing at softgen.ai

Edge: Emerging tool. Check free tier status before committing.

Link: softgen.ai

23. Create.xyz

Design-to-code + full-stack generation.

Pricing: Verify current pricing at create.xyz

Edge: Blends design and code generation. Good for designers crossing into dev.

Link: create.xyz

24. Anthropic API (Claude)

Home of Opus 4.5, Sonnet 4.5, Haiku 4.5. Best reasoning in the market.

Pricing: $5 free signup · $25K startup credits · Pay-as-you-go after

Edge: Prompt caching = 90% savings on repeats. Batch API = 50% off. Stack both = 95% savings.

Link: console.anthropic.com

25. OpenAI API

GPT-4o, GPT-4.1, o3. The default for most teams.

Pricing: $5 free signup · $2,500 startup · $25K open-source · Pay-as-you-go

Edge: Largest ecosystem of integrations. Fastest general-purpose responses.

Link: platform.openai.com

26. Google AI Studio (Gemini)

Cheapest frontier model API, period.

Pricing: Free tier (15 RPM) · Gemini 1.5 Flash $0.075/1M input

Edge: Gemini Flash is the cheapest capable model. Great for bulk processing.

Link: aistudio.google.com

27. Groq

LPU-based inference. 10-20x faster than GPU providers.

Pricing: FREE forever (rate-limited) · Pay for more throughput

Edge: Insanely fast. Perfect for real-time apps. No credit card to start.

Link: groq.com

28. Together AI

Hosted open-source models (Llama, DeepSeek, etc.).

Pricing: $25 new user credits · Pay-as-you-go after

Edge: Best quality/price on open models. Good fallback for expensive closed models.

Link: together.ai

29. Fireworks AI

Fast, cheap open-model hosting. Competitive with Together.

Pricing: $1 signup · 10 RPM free · 6,000 RPM with payment

Edge: Fine-tuning is easy. Great for custom model deployment.

Link: fireworks.ai

30. OpenRouter

Unified API routing. Auto-picks the cheapest available model.

Pricing: Free (BYO model keys, zero markup)

Edge: Set fallback chains across providers. The cost-optimizer's secret weapon.

Link: openrouter.ai

RULE #1: APPLY FOR EVERYTHING

Applying to these programs takes minutes and costs nothing. The rejection rate is low if you have a real project or a valid .edu email. Stop overthinking it.

PROGRAM

WHO

WHAT YOU GET

LINK

GitHub Student Dev Pack

.edu email

Copilot Pro free + JetBrains + more

education.github.com/pack

Cursor Student

.edu email

1 YEAR of Pro FREE ($240 value)

cursor.com/students

JetBrains Student

.edu email

All IDEs free (via GitHub Pack too)

jetbrains.com/student

Vercel Pro for Students

.edu email

Pro tier free for 1 year

vercel.com/for/students

Azure for Students

.edu email

$100 credit + free OpenAI services

azure.microsoft.com/students

Google Cloud Student

.edu email

$200-300 credit for Vertex AI / Gemini

cloud.google.com/edu

GitHub Copilot Free

Any user

2,000 completions + 50 chats / month

github.com/features/copilot

Supermaven Free

Any user

Unlimited autocomplete FOREVER

supermaven.com

Continue.dev

Any user

100% free, BYO API key

continue.dev

Warp Free

Any user

75 AI credits/mo (150 first 2 months)

warp.dev

Groq Free

Any user

Forever free, rate-limited

groq.com

PROGRAM

WHO

WHAT YOU GET

LINK

Anthropic API

New signup

$5 free (3-month expiry)

console.anthropic.com

OpenAI API

New signup

$5 free (3-month expiry)

platform.openai.com

Google AI Studio

New signup

Free tier (15 RPM ongoing)

aistudio.google.com

Together AI

New signup

$25 free credits

together.ai

Fireworks AI

New signup

$1 credit + free rate-limited

fireworks.ai

Groq

New signup

Free forever, no card required

groq.com

Replicate

New signup

Free tier + $1 credit

replicate.com

PROGRAM

WHO

WHAT YOU GET

LINK

Anthropic Claude for Startups

Pre-seed/seed founders

Up to $25K credits

claude.com/startups

OpenAI Startup Program

Pre-Series B startups

$2,500 + deep discounts

openai.com/startups

OpenAI Researcher Access

University affiliation

$1,000 credits

openai.com/research

OpenAI Codex Open Source Fund

OSS maintainers

Up to $25K credits

openai.com/index/codex

AWS Activate

Pre-Series B

$1K - $300K depending on stage

aws.amazon.com/startups

Google for Startups Cloud

Pre-Series A

$2K - $350K over 2 years

cloud.google.com/startup

Azure for Startups (Founders Hub)

Any founder

$1K - $150K credits

microsoft.com/startups

Together AI Startup

Pre-seed/seed

Up to $50K credits

together.ai/startup

Notion for Startups

Via accelerator partner

6 months Plus free

notion.com/startups

THE HACKATHON LOOP

Every 2-3 weeks, Anthropic / OpenAI / Google drop a hackathon with free credits as participation rewards. Typical drops:

• Anthropic: $100-500 in API credits per participant
• OpenAI: $100-300 in API credits + GPT access
• Google: $200-500 in Vertex AI + Gemini credits
• Groq / Fireworks / Together: $50-200 per event

01. Prompt caching (Claude API)

Cache your system prompt, docs, and reference files. Writes cost 1.25x (5-min TTL), reads cost 0.1x. Pays off after ONE read. Up to 90% savings on repeat input tokens.

02. Batch API (50% off)

Anthropic and OpenAI both offer Batch APIs with 50% discount on input AND output. Use for anything that doesn't need real-time: bulk analysis, doc processing, commit message generation.

03. Stack Cache + Batch (95% savings)

Combine prompt caching with the Batch API. Cached read at 0.1x + 50% batch discount = up to 95% off your input bill. This is the single biggest unlock in the playbook.

04. Model tiering (Haiku → Sonnet → Opus)

Haiku for file naming, commit messages, simple edits. Sonnet for mid-weight coding. Opus ONLY for hard debugging or architecture decisions. Haiku is ~4x cheaper than Sonnet.

05. CLAUDE.md / .cursorrules files

Write your conventions, architecture, and style guide once. Every new chat inherits it automatically. Stop retyping "use TypeScript, prefer functional components" every session.

06. Subagent pattern (context hiding)

In Claude Code, spawn a subagent for exploration or verification. Subagent eats the tokens, returns a summary. Your main context stays clean. Cuts main-session token burn by 40-60%.

07. @-mention specific files, never "@repo"

Dumping a whole repo into context is the #1 way people burn tokens. Use @file.ts or @folder/ instead. Saves 30-70% per request.

08. Clear chat after every task

Don't let conversations grow to 200 messages. Start a new chat for every new feature. Long threads have exponential token cost.

09. Use Groq for low-stakes drafts

Brainstorming, first drafts, simple transforms? Use Groq's free Llama 3.3 70B. It's 20x faster than Claude and costs you zero.

10. OpenRouter fallback chains

Configure: Claude Sonnet → GPT-4o → DeepSeek. If the expensive one is rate-limited or down, auto-fallback to cheaper. Zero markup.

11. Autocomplete FREE, chat PAID

Supermaven (unlimited free) or Copilot Free (2K/mo) for autocomplete. Save expensive chat-mode for debugging and architecture. Most devs get 60% of value from autocomplete alone.

12. Aider with --cache-prompts

Aider has built-in Claude prompt caching. Add the flag and run --cache-keepalive-pings 12 to keep the cache alive for 1 hour between prompts. 90% cheaper repeat requests.

13. Disable always-on AI when reading

Cursor Tab and Copilot burn tokens constantly. Toggle off when you're reading code or navigating. Re-enable when you start writing.

14. Local models for trivial stuff

Ollama + Qwen 2.5 Coder 7B (4GB RAM) handles boilerplate, HTML scaffolding, simple refactors. Runs on your MacBook. Zero API cost, zero latency.

15. Precompute embeddings locally

Don't pay OpenAI for embeddings on your own docs. Use Ollama's nomic-embed-text locally. One-time cost of 2 minutes of GPU, saves ~$20/mo for a mid-size codebase.

16. Prompt compression

Rewrite your system prompt in ultra-terse form. Remove examples that aren't load-bearing. Use bullets instead of prose. Cuts input tokens 20-40%.

17. Cloudflare AI Gateway

Wraps your API calls with automatic caching + observability. Free tier generous. Cache identical requests = free responses. Bonus: you get analytics for free.

18. Gemini Flash for bulk processing

Flash costs $0.075/1M input — roughly 40x cheaper than Sonnet. For classification, tagging, summarization tasks, it's embarrassingly good enough.

19. Don't pay for what you get free

Every creator uses $20 Cursor. Very few apply for their $240 student year. Very few claim $5 Anthropic signup credits. Very few claim $25 Together signup. Free money, left on the table.

20. Windsurf > Cursor if you're price-sensitive

Windsurf Pro ($20) has unlimited SWE-1.5 agent requests. Cursor Pro ($20) has rate-limited agents. Same price, different ceilings. Test both before picking.

STACK 1 — BROKE STUDENT

Target spend: $0/month. Target capability: 80% of paid stacks.

IDE: Cursor Hobby OR VS Code + Continue.dev

Autocomplete: Supermaven Free (unlimited) or Copilot Free (2K/mo)

Chat model: Claude Code CLI on $5 Anthropic signup credits

Vibe coding: Bolt.new free (300K tokens/day) + v0 daily bonus

Terminal: Warp Free (75 credits/mo)

Speed API: Groq Free (Llama 3.3 70B, unlimited speed)

Claim: GitHub Student Pack + Cursor 1 year free + Vercel Pro free + Azure $100

Monthly cost: $0

STACK 2 — SOLO FOUNDER

Target spend: $40-50/month. Target capability: ship faster than a 5-person team.

IDE: Windsurf Pro ($20/mo) — unlimited SWE-1.5 agents

Chat: GitHub Copilot Pro ($10/mo) — multi-model

Heavy lifting: Claude Code CLI + Anthropic API ($10-20/mo with caching)

Vibe coding: v0 for components + Bolt.new free tier for scaffolds

Cost control: OpenRouter routing + prompt caching + Batch API

Claim: $25K Anthropic Startup + $2.5K OpenAI Startup + AWS Activate

Monthly cost: $40-50 (net ~$0 first 6 months with startup credits)

STACK 3 — POWER USER

Target spend: $200-500/month. Target capability: no ceiling.

Primary IDE: Cursor Pro+ ($40/mo) + Claude Code Max ($100-200/mo)

Secondary IDE: Google Antigravity (free) for parallel agent tasks

Autocomplete: Supermaven Pro ($10/mo)

Vibe coding: Bolt Pro + v0 Premium + Lovable Pro

API budget: Anthropic + OpenAI pay-as-you-go with caching enabled

Terminal: Warp Build ($15/mo)

Monthly cost: $200-500 gross · $100-250 net with caching/batching

WHICH STACK IS RIGHT?

• Student with no income → Stack 1. Period. You can build a real startup on this.
• Bootstrapped founder → Stack 2. Apply for startup credits to cover months 1-6.
• Funded team → Stack 3. Your time is the bottleneck, not your tool budget.

01. Claim the GitHub Student Pack (3 min)

Go to education.github.com/pack. Verify with your .edu email. Unlocks Copilot Pro, JetBrains, Vercel, and 100+ other tools. This is the single highest-ROI click in the entire doc.

02. Activate Cursor Student (2 min)

Go to cursor.com/students. Verify .edu. Get 1 year of Pro free. Nothing else to do.

03. Create Anthropic + OpenAI + Groq accounts (5 min)

Sign up at console.anthropic.com, platform.openai.com, and groq.com. You now have $10 in signup credits plus unlimited free Groq inference. Save the API keys.

04. Install Continue.dev or Cline in VS Code (3 min)

Open VS Code extensions. Search "Continue" or "Cline". Install. Paste your Anthropic + Groq keys. You now have a fully-featured AI IDE for $0.

05. Install Supermaven + Warp (4 min)

Supermaven: free unlimited autocomplete. Warp: free AI terminal. Both take 2 minutes to install. Your daily driver tools are now upgraded.

06. Apply for startup credits (10 min)

If you have ANY project idea, apply to Anthropic Startups, OpenAI Startup Program, AWS Activate, Google for Startups, and Azure Founders Hub. Even rough-draft projects get accepted regularly.

07. Set up a .cursorrules or CLAUDE.md file (3 min)

In your main project, create CLAUDE.md. Write your architecture, style guide, and preferences. Every AI session from now on will inherit this context. Immediate token savings.

YOUR 30-MINUTE HOMEWORK

1. Claim the GitHub Student Pack

2. Claim Cursor 1-year student Pro

3. Sign up for Anthropic + OpenAI + Groq

4. Apply to every startup credit program you qualify for

5. Write your CLAUDE.md / .cursorrules file

ENJOYED THIS TOOLKIT?

This is the kind of stuff I publish every week in the Tensor Protocol — zero-fluff breakdowns of every tool, trick, and workflow in the AI space.

Follow @tensor._.boy on Instagram