Free Artificial Intelligence: Tools, Courses, and Real-World Uses in 2025
This guide to free artificial intelligence in 2025 shows you the best no-cost AI tools, courses, and real-world uses. Whether you’re a student, educator, developer, or small business owner, this guide will help you find the best free AI resources to boost your productivity and learning. Understanding free AI options matters more than ever: students can accelerate their studies, professionals can automate repetitive tasks, and small businesses can innovate without breaking the bank. We’ll cut through the noise to show you exactly what’s worth your time-no ads, no filler, just the signal.
Key Takeaways
Major providers including Google, OpenAI, Meta, Microsoft, and Hugging Face (a hub for accessing pre-trained models and datasets, supporting NLP, computer vision, and audio – see fact 5) all offer always-free tiers for their AI models, though each comes with specific limits on usage, features, or geographic availability.
You’ll discover concrete examples throughout this article: specific free tools for chat, code, images, and video; named courses with exact durations (like 5–6 hour introductory programs); and practical ideas such as turning spreadsheets into websites, restoring old family photos, or learning async JavaScript.
This article reflects KeepSanity AI’s philosophy: focus only on the most impactful, widely available free AI options and use cases that actually matter for your work and life.
The FAQ section at the end addresses common concerns about safety, data privacy, and whether “free” really means free-questions that matter before you upload your first document.
Always-Free AI Tools and Platforms: The Essentials
Here are the most important always-free AI tools and platforms you should know about:
Google Gemini: Free to use for users 18 and over with a personal Google Account or a Google Workspace account (fact 20). Handles text, images, and code.
Google AI Studio: Allows users to prototype and run prompts in their browser, making it accessible for developers, students, and researchers (fact 21, 19).
Hugging Face: A hub for accessing pre-trained models and datasets, supporting NLP, computer vision, and audio (fact 5).
NotebookLM: A free tool for summarizing uploaded source material and surfacing insights between topics (fact 17, 2).
Google Colab: Provides free access to cloud GPUs, allowing researchers to run machine learning workloads (fact 3).
Kaggle: Offers free GPU resources, datasets, and a collaborative coding environment (fact 4).
GitHub Copilot: Provides context-aware code suggestions in major IDEs (fact 6).
Canva Magic Design: AI features like Magic Design help generate graphics and presentations from simple prompts (fact 7).
Keras: A high-level, user-friendly API that runs on top of TensorFlow, ideal for fast prototyping (fact 8).
TensorFlow: A production-grade platform offering high scalability and a comprehensive ecosystem for deploying models (fact 9).
Scikit-learn: The primary library for classical machine learning tasks like classification and regression in Python (fact 10).
Leonardo.ai and DALL-E 3 (via Bing/Microsoft): Image creation tools (fact 11).
Perplexity AI: Provides direct, cited answers to research questions (fact 12).
Semantic Scholar: A free, AI-powered academic search engine (fact 13).
Research Rabbit: A tool for mapping literature citations (fact 14).
Connected Papers: A visual tool for mapping relationships between academic papers to find similar research (fact 15).
Brisk: An AI platform that personalizes learning materials and adapts to student needs within existing tools like Google and Microsoft (fact 16).
What “Free Artificial Intelligence” Really Means in 2025
Free artificial intelligence in 2025 refers to access to AI models, applications, APIs, and educational resources without upfront monetary costs. This includes everything from completely unrestricted personal use to tiered offerings with defined quotas that reset monthly.
The differences between use cases matter:
Personal use: Generally the most generous free tiers, designed for individuals experimenting or learning
Academic use: Often includes extended access for verified students and educators
Commercial limits: Many free tiers restrict business applications or require attribution
Key constraints you’ll encounter across free AI tools:
Rate limits: Most services cap how many requests you can make per minute (e.g., 15 requests per minute for some models)
Monthly quotas: Translation APIs might offer 500,000 characters, Vision APIs around 1,000 image analyses, Speech-to-Text approximately 60 minutes
Model versions: Free users typically get “Flash” or “mini” editions with reduced context windows (128K tokens versus 1M tokens on paid plans)
Geographic availability: Some features remain unavailable in the EU due to GDPR requirements, while others vary by region
Big-cloud providers like Google Cloud, Azure, and AWS separate their free tier usage from paid tiers through billing dashboards. Staying within these limits results in no charges-but crossing them triggers pay-as-you-go billing. Always enable billing alerts before you begin experimenting.
The rest of this article focuses on tools that are genuinely usable without entering a credit card first.
Work with the Latest AI Models Using Free AI Tools
As of 2025, users can access cutting-edge models like Google Gemini 2.0 Flash, GPT-4o mini, Meta Llama 3, and Mistral models for free through web apps and limited APIs. These aren’t watered-down demos-they’re capable systems handling real work.
Free Web Frontends for AI Models
Concrete free web frontends you can use today:
Google Gemini (gemini.google.com): Access Gemini 2.0 Flash for reasoning, coding, translation, and multimodal tasks
ChatGPT Free (chatgpt.com): Runs GPT-4o mini with image uploads and voice mode
Microsoft Copilot (copilot.microsoft.com): Integrates Bing search with GPT-4o mini capabilities
Perplexity Free Tier: Offers cited research synthesis for quick fact-checking
What These Free AI Tools Can Handle
Translate content across 100+ languages in real-time
Generate images from text descriptions (with daily limits around 50 high-res outputs)
Plan and analyze videos by storyboarding scripts
Recognize speech with 95%+ accuracy on clear audio
Debug code by refactoring JavaScript patterns or explaining complex functions
For students, indie developers, and small businesses, these capabilities cover roughly 80-90% of individual workloads without spending a dollar.
Which AI Tools Are Actually Free to Use Long-Term?
The difference between free trials and permanent free tiers matters when you’re building workflows you can rely on. Many services offer introductory credits that expire, while others maintain ongoing free access indefinitely.
Tools with permanent free access as of early 2025:
Google Gemini: Free for users 18+ with personal Google Accounts, handles text, images, and code
Google AI Studio: Free tier for Gemini API prototyping with lower rate limits
OpenAI ChatGPT Free: Runs on GPT-4o mini with multimodal capabilities
Hugging Face Spaces and Inference Endpoints: Free hosting for demos and model inference on Llama 3 (8B/70B parameters)
Meta Llama 3: Accessible via third-party UIs like Perplexity without direct costs
Google Cloud AI APIs with non-expiring monthly quotas:
Translation API: ~500,000 characters (roughly 100,000 words)
Vision AI: ~1,000 units for image analysis
Speech-to-Text: ~60 minutes of audio
Text-to-Speech: ~4 million characters for Standard voices
Natural Language: ~5,000 entity analyses
These quotas support hundreds of hours of light media processing monthly-ideal for class projects, prototypes, or small internal tools. Note that quotas and terms may change, so verify current limits on provider pricing pages before building dependencies.
Is There a Single “Free AI Website” Directory Worth Bookmarking?
Most major providers maintain public directories of their free and paid AI tools, but no single “complete” directory exists that covers everything.
Worth bookmarking:
Google’s AI product directory: Consolidated view of Gemini, Vertex AI, and related services
Hugging Face’s model hub: Over 1 million models with free downloads and demo Spaces (Hugging Face is a hub for accessing pre-trained models and datasets, supporting NLP, computer vision, and audio – see fact 5)
Awesome-LLM lists on GitHub: Community-curated open-source model collections
AIxploria: Catalogs 5,000+ free tools across categories
Rather than chasing every new launch, build a personal shortlist of 3–5 tools that cover your core needs. User studies show 70% abandonment rates when people try to track every new release.
KeepSanity’s weekly newsletter serves as a curated pointer to important new free tools rather than an exhaustive directory-covering only the shifts that actually matter.
Free Generative AI for Text, Chat, and Code
Generative AI is now accessible for zero cost through browser-based chat interfaces and low-cost APIs from major cloud platforms. The barrier to building with these models has never been lower.
Free Generative AI Platforms
Google’s Vertex AI offers:
Low per-token pricing (around $0.0001 per 1,000 characters of context)
$300 in credits for new accounts
Permanent free tier for select models
Enough capacity to prototype chatbots processing millions of tokens monthly
Microsoft Azure OpenAI and OpenAI’s API provide similar on-ramps:
Introductory credits (typically 1M tokens free)
Modest free calls for GPT-4o mini
Ideal for prototyping but not heavy production workloads
Developers can build chatbots, code assistants, and small apps on these free tiers. You only need to upgrade when crossing defined monthly limits-which most individual projects never do.
Free AI Writing and Coding Assistants
Many “AI writing tools” are thin wrappers around large models that themselves offer free access. Before paying for a wrapper, check if you can use the underlying model directly.
Concrete free options:
Gemini in Google Docs and Gmail: Draft emails and documents directly in Google Workspace
GitHub Copilot: Provides context-aware code suggestions in major IDEs (fact 6). Free for verified students and open-source maintainers
Gemini Code Assist: Preview access for code completion
Replit Ghostwriter Free Tier: Debug Python and JavaScript with explanations
VS Code Extensions (Continue.dev): Use local or GPT-4o mini backends
Common use cases that yield 2-3x productivity:
Drafting professional email summaries from bullet notes
Refactoring 100-line JavaScript from callbacks to async/await
Generating documentation from code comments
Creating test cases from function signatures
Limitations include context caps (4K-32K tokens on free tiers) and no enterprise data isolation.
Are Flagship Models Like Google Gemini or GPT-4o Really Free?
Flagship models usually have both free and paid versions with meaningful differences in limits, speed, and advanced features.
Google Gemini:
Free for users 18+ with personal Google Accounts
Runs Gemini 2.0 Flash (optimized for speed with reduced parameter counts)
Optional paid Gemini Advanced unlocks 1M token context and priority access
Workspace access depends on admin settings
OpenAI ChatGPT Free:
Runs on GPT-4o mini as of 2025
Includes multimodal capabilities (image uploads, basic file handling up to 10MB)
Queues during peak usage
Lacks custom GPTs and enterprise features
These free access paths are ideal for individuals learning or experimenting but insufficient for high-traffic commercial apps needing 99.9% uptime.
Transition:
With these tools, you can handle most text, chat, and code tasks for free. Next, let’s explore how AI agents and automation can further streamline your workflows.
Free AI Agents and Automation: What You Can Do Without Paying
AI agents are systems that autonomously call tools, APIs, and workflows-reading documents, then triggering email sequences, for example. Limited agent building is now possible on free tiers.
Free AI Agent Platforms
Google’s Vertex AI Agent Builder:
Free tier supports prototyping with low limits (~100 agent runs/month)
“Agent Garden” blueprints for FAQ bots scanning PDFs
Suitable for classroom use and early experiments
Other free options:
Microsoft Copilot Studio trials: Build lead qualification flows
LangChain: Run with locally hosted or free-tier models
AutoGen and LangGraph: Multi-agent simulations on community Colab GPUs (Google Colab provides free access to cloud GPUs, allowing researchers to run machine learning workloads – fact 3)
Sample automations you can build without paying:
FAQ bots for a small site handling ~1,000 queries monthly
Knowledge base search across 50 company PDFs
Simple lead-qualification flows from form submissions
Internal document summarizers for team meetings
Scaling hits quotas quickly, but prototyping costs nothing.
Free Cloud AI Quotas: How Far Can You Go at Zero Cost?
Google Cloud, AWS, and Azure all provide free usage each month across bundled AI and non-AI services. Staying within limits means no charges.
Google Cloud free monthly quotas:
Service | Free Quota | Approximate Use |
|---|---|---|
Translation API | 500,000 characters | ~100,000 words |
Vision AI | 1,000 units | 1,000 image analyses |
Speech-to-Text | 60 minutes | ~1 hour of audio |
Text-to-Speech (Standard) | 4 million characters | Hours of synthesized speech |
Natural Language | 5,000 units | Entity/sentiment analyses |
AWS Free Tier offers equivalents like 5 million Rekognition image operations annually. Azure provides approximately 20,000 Speech transactions.
These quotas cover experiments, class projects, or small internal tools-but cap at roughly 1-5% of production-scale requirements. Always check your billing dashboard to avoid surprises.
Transition:
Beyond automation, free AI is also revolutionizing creative work-enabling anyone to generate images, videos, and designs at no cost.
Free AI for Creative Projects: Images, Video, and Design
Creative AI tools for imaging, video, and design exploded in 2024-2025. Many are capped by resolution, watermarks, or daily generation limits-but they’re still powerful enough for real work.
Free Creative AI Tools
Google ecosystem tools:
Gemini with Imagen 3: High-res image generation from text or sketches (daily limits ~100 images)
Veo 3.1: 1080p videos up to 60 seconds (monthly quota ~10-20 clips)
Canvas: Interactive app prototypes from Sheets data
Non-Google options:
Canva Magic Studio Free: Thumbnails and social graphics (unlimited low-res)
Adobe Express Free Tier: Generative fill with ~50 credits monthly
Stable Diffusion via Hugging Face Spaces: Free GPU time for hours daily (Hugging Face is a hub for accessing pre-trained models and datasets, supporting NLP, computer vision, and audio – see fact 5)
Use free creative AI for thumbnails, social posts, product labels, and concept art. High-volume commercial campaigns-where licensing and quality become critical-warrant paid plans.
Transformation Workflows: From Sketches to Finished Assets
Many free tools now focus on specific transformations: turning one artifact into something polished.
Concrete transformation workflows (each under 30 minutes):
Hand-drawn wireframe → realistic UI mockup: Upload a sketch to Gemini Image models, prompt “polished Figma style,” receive a professional-looking interface design
Google Sheets → simple landing page: Export data as JSON, use Canvas-style tools to generate HTML/CSS
Old family photo → restored colorized version: Remove noise and inpaint missing details via Vision API, add era-appropriate color
Notes → web app mockup: Paste bullet points, request a working prototype layout
Some tools mark outputs as “AI-generated” or apply small watermarks on free plans-check before using for client work.
Text-to-Video and Vertical Video Creation on a Budget
Text-to-video and vertical formats (Reels, TikTok, Shorts) are now reachable via free tiers, though usually with short durations and platform branding.
Veo 3.1 and similar tools:
Sync AI-generated voices with mouth movements
Require Google sign-in but not paid subscription
Support vertical formats optimized for mobile
Other free options:
Tool | Free Limits | Typical Constraints |
|---|---|---|
Pika Labs | 100 credits/month | 720p export |
Runway Trial | 125 credits | Watermark included |
CapCut AI Templates | Unlimited low-fi | Template-based only |
Start with one tool and iterate rather than chasing every new video model release.
Transition:
Beyond creative work, free AI is also transforming how we learn and develop new skills.
Free AI for Learning: Courses, Study Aids, and Career Growth
Beyond tooling, free AI education has exploded. High-quality courses from 2023–2025 and integrated AI tutors make developing skills more accessible than ever.
Google’s Free AI Courses
AI Essentials: 5 hours covering basics and responsible use
Prompting Essentials: 6 hours teaching the 5-step prompting method
AI for Educators: 2 hours focused on lesson planning
AI modules in Google Career Certificates: Embedded throughout various programs
IBM SkillsBuild
IBM SkillsBuild offers free artificial intelligence courses for students covering:
AI basics and data science fundamentals
NLP and machine learning ethics
Hands-on chatbot and robot projects
These offerings prepare learners for a rapidly growing job market. AI mentions in job postings grew over 120% between 2023–2025, and AI skills correlate with 25-40% salary premiums.
Essential AI Skills Everyone Should Learn
“AI literacy” in 2025 includes prompt design, understanding basic model capabilities and limitations, and recognizing data privacy and bias concerns.
Core learning outcomes worth developing:
Structure prompts in 5 clear steps: set context, define task, specify format, give examples, refine output
Create a reusable library of prompts for recurring work tasks
Summarize documents, analyze data, and draft content using AI tools
Spot hallucinations and verify factual claims independently
Understand when AI outputs need human review
Free routes to these competencies:
Google’s AI skill tracks on Skillshop
Short Coursera modules (Andrew Ng’s 11-hour GenAI course)
University-backed open courses (MIT, Stanford CS221 lectures on YouTube)
Using AI to Accelerate Your Job Search
Free AI can function as a 24/7 career coach-helping identify transferable skills, rewrite resumes, track applications, and simulate interviews.
Courses like “Accelerate Your Job Search with AI” (approximately 6 hours) teach how to use Gemini and NotebookLM (a free tool for summarizing uploaded source material and surfacing insights between topics – see fact 17) to:
Research target companies and roles
Practice behavioral interview questions
Build a structured job search plan
Specific workflows you can implement:
Generate target-company briefs: Paste a job posting, ask AI to summarize the company’s recent news, culture indicators, and likely interview topics
Tailor cover letters: Upload your resume and job description, request a customized letter highlighting relevant experience
Create interview prep flashcards: Generate spaced-repetition questions from common behavioral prompts
The goal is landing interviews faster and negotiating better-AI handles the research grunt work so you can focus on performance.
Prompting Essentials and AI Fundamentals
Practical prompting skills dramatically improve output quality from any free model. This makes prompting a foundational skill worth developing early.
Google’s 5-step prompting approach:
Set context: “As a JavaScript expert…”
Define task: “…explain async/await…”
Specify format: “…with an event loop analogy…”
Give examples: “…include this code snippet: [paste code]…”
Refine: “…simplify the explanation for a beginner”
AI fundamentals courses (typically 4–6 hours) cover:
Generative AI basics and model types (LLMs, vision, speech)
Responsible use principles
Enough understanding to evaluate tools rather than using them blindly
Complete at least one short fundamentals course and one prompting course to maximize value from every free AI tool in this article.
AI for Students, Educators, and Small Businesses
Different audiences get different value from free AI. Here’s how each group can integrate these tools:
For students:
Create study guides, flashcards, and custom quizzes from lecture notes
Use NotebookLM-style tools to summarize PDFs, websites, and videos with audio overviews
Generate practice problems for exam preparation
Get explanations of complex concepts in simpler terms
For educators:
Tools like Brisk (an AI platform that personalizes learning materials and adapts to student needs within existing tools like Google and Microsoft – see fact 16) integrate with Google Docs and LMS platforms
Generate differentiated materials for varied learning levels with personalized instruction
Create rubrics and provide feedback more efficiently
Build interactive lessons and teach with AI assistance
For small businesses:
Courses like “AI for Small Businesses” cover practical applications
Free tools help with copywriting, basic analytics, and customer support chatbots
Simple automation handles repetitive tasks
Prototype ideas before investing in paid solutions
Transition:
With these learning resources and skills, you can unlock new opportunities in your studies and career. Next, let’s look at practical ways to use free AI in everyday life.
Practical Ways to Use Free AI in Everyday Life
Beyond work and study, free AI can streamline personal tasks: travel planning, budgeting, hobbies, creative projects, and organization.
Trip Planning
Trip planning: Generate detailed itineraries with logistics and local tips
Ask Gemini, Copilot, or ChatGPT to draft comprehensive itineraries. A 10-day Greece trip prompt might yield:
Day-by-day Athens, Mykonos, and Santorini schedules
Ferry booking tips and seasonal considerations
Restaurant recommendations by neighborhood
Budget breakdowns for different travel styles
Combine AI text with other tools:
Plot locations on Google Maps for visual planning
Generate image previews of destinations
Create hiking-focused guides for US national parks
Example itineraries worth generating:
3-day family trip to San Diego with kid-friendly activities
Art-and-history tour through Florence, Kyoto, or Prague
Weekend hiking adventure in nearby state parks
Still verify critical details manually: opening hours change, visa requirements vary, and safety advisories update frequently.
Career Pivots
Career pivots: Create structured transition plans with milestones
Planning a 6-month transition into software testing, UX design, or data analysis? Free AI helps structure the journey.
How AI assists career changes:
Break goals into monthly milestones with specific deliverables
Generate budget templates for transition periods
Suggest learning resources and project ideas in your target field
Identify transferable skills from your current role
Side project and business planning:
Draft business plans for a board-game shop, coffee cart, or online course
Cover executive summary, market analysis, marketing strategy, and financial projections
Test assumptions by asking AI to critique your plans
AI offers structure and starting points. Real-world validation through conversations with industry professionals and potential customers remains essential.
Learning Technical Topics
Learning new skills: Get explanations of complex topics in approachable terms
Use asynchronous JavaScript as an example. AI can explain the event loop, callbacks, Promises, and async/await with analogies that stick.
Sample workflow:
Ask: “Explain JavaScript’s event loop like I’m a restaurant manager coordinating waiters”
Get analogy: The event loop is like a restaurant manager-orders (callbacks) queue up, the manager dispatches waiters (execution) one at a time, and customers (the browser) stay responsive while waiting
Paste your own code: “Refactor this callback pattern to use async/await”
Request test generation: “Write unit tests for this function”
This technique generalizes across domains:
Explaining Kubernetes manifests in plain English
Breaking down SQL queries step by step
Demystifying machine learning model architectures
Free tiers handle individual debugging and learning sessions easily.
Creative Writing, Games, and Personal Hobbies
Creative writing: Brainstorm ideas, develop characters, overcome blocks
Free AI generates story prompts, game ideas, character backstories, and creative exercises for classes or personal projects.
Concrete examples:
Brainstorm short story ideas by genre for a creative writing class
Generate a 90s grunge or 2000s pop persona description from an uploaded photo
Design trivia quiz questions from audio notes or podcast episodes
Create character sheets for tabletop RPG campaigns
Low-cost entertainment ideas for students:
AI-generated scavenger hunt clues for campus events
Personalized movie recommendation lists based on mood
DIY escape room puzzle concepts
Party game questions tailored to your friend group
Maintain your own voice-AI provides starting points and ideas, not finished creative work.
Transition:
With these practical applications, free AI can enhance your daily life, work, and learning. To stay ahead, it’s important to keep up with the best new tools-without getting overwhelmed.
How KeepSanity Helps You Track the Best Free AI – Without the Noise
KeepSanity AI delivers one email per week covering only the major AI news that actually happened. No daily filler. Zero ads. Just the signal.
Why most AI newsletters waste your time:
Daily emails exist to tell sponsors “our readers spend X minutes per day with us”
Content gets padded with minor updates that don’t matter
Sponsored headlines you didn’t ask for burn your focus
What KeepSanity provides instead:
Coverage across business, product updates, models, tools, resources, robotics, and trending papers
Smart links (papers link to alphaXiv for easy reading)
Scannable categories for skimming everything in minutes
Alerts when major platforms change free tiers, launch new no-cost tools, or add impactful AI courses
Subscribed by teams like Adobe, Surfer, and Bards.ai-professionals who need to stay informed but refuse to let newsletters steal their sanity.
Lower your shoulders. The noise is gone. Here is your signal.
FAQ: Free Artificial Intelligence in 2025
Is free AI really safe to use with my personal or business data?
Safety depends entirely on provider policies. ChatGPT and Gemini use free-tier inputs to improve models by default, though opt-out settings exist. GDPR-compliant modes operate in relevant regions.
Practical guidance:
Avoid pasting highly sensitive data (trade secrets, medical details, unreleased financials) into consumer-grade free tools
Check data use settings and privacy policies before building workflows
Use enterprise offerings with clear data-processing agreements when confidentiality is critical
Free AI works fine for everyday tasks-just not as a replacement for secure, audited systems
Can I build a serious business on free AI tiers alone?
Free tiers excel for prototyping, learning, and early experiments-but they’re intentionally limited so sustained or high-volume use requires payment.
Recommended approach:
Prototype on free tiers and measure approximate usage (tokens, images, minutes)
Price out paid plans before scaling up
Budget for at least one paid model or platform once you reach consistent user traffic
Consider vendor lock-in and export options from day one
Treat AI usage as an operational cost similar to hosting or email infrastructure, not as a permanently “free” resource.
How do I keep up with new free AI tools without getting overwhelmed?
Cap yourself at a small personal stack-one main chat model, one creative tool, one coding assistant. Revisit choices quarterly instead of chasing every launch.
Staying informed without burnout:
Use curated sources like KeepSanity’s weekly email for major shifts only
Create an “AI experiments” document logging tools tried, what worked, and what to drop
Resist the urge to sign up for every new platform
The goal is to ship work and develop skills, not collect accounts.
What skills should I learn first to get the most from free AI?
Start with three core skills:
Writing clear prompts (take Google’s “Prompting Essentials” course)
Basic critical thinking about AI outputs (spotting hallucinations, verifying facts)
One domain-specific application (coding, writing, data analysis)
Building proficiency:
Complete short courses like “AI Essentials” (5 hours) and one fundamentals course
Practice by applying AI daily: summarizing meetings, drafting emails, generating study questions
Skillful users extract far more value from the same free tools than casual dabblers
Practice matters more than chasing the newest model.
Will free AI disappear as models get more expensive to run?
Some form of free access is likely to remain. Providers rely on free tiers to attract users, gather feedback, and maintain market share against competitors.
Reasons for optimism:
Models are becoming more efficient (inference costs dropped 10x between 2023-2025)
Open-source ecosystems continue lowering barriers to running capable models locally
Competition among providers keeps free options alive
Expect details to change (quotas, model variants), but the pattern of “usable free access + paid upgrades” should persist. Subscribe to curated updates to catch major shifts when they happen.