ChatGPT (OpenAI) ▶ Watch
OpenAI released ChatGPT in November 2022, sparking the current wave of public interest in AI. ChatGPT is powered by OpenAI's GPT family of models (GPT stands for "Generative Pre-trained Transformer"), including reasoning-focused models. It is known for its very large user base, a broad set of built-in features - image generation, voice conversation, file analysis, web search, custom assistants - and a large ecosystem of developers using the OpenAI API. Microsoft is a major partner and investor, and OpenAI models power several Microsoft Copilot products.
Claude (Anthropic) ▶ Watch
Anthropic was founded in 2021 by a group of former OpenAI researchers with a strong focus on AI safety research. Its assistant, Claude, comes in model tiers that trade off capability, speed and cost - historically named Opus (most capable), Sonnet (balanced) and Haiku (fastest). Anthropic developed Constitutional AI, a training approach that guides the model with a written set of principles. Claude is widely used for long-document analysis, careful writing, and software development, including agentic coding tools such as Claude Code.
Gemini (Google) ▶ Watch
Google's assistant was launched as Bard in 2023 and renamed Gemini in early 2024, after the Gemini model family built by Google DeepMind. Gemini models were designed from the start to be natively multimodal - handling text, images, audio and video - and come in tiers such as Pro (most capable) and Flash (fast and economical), plus small Nano models that run on devices. Gemini's biggest advantage is integration with Google's products: Search, Gmail, Docs, Drive, YouTube and Android.
Side-by-side comparison ▶ Watch
| ChatGPT | Claude | Gemini | |
|---|---|---|---|
| Company | OpenAI | Anthropic | Google / Google DeepMind |
| Model family | GPT and reasoning models | Claude (Opus, Sonnet, Haiku tiers) | Gemini (Pro, Flash, Nano tiers) |
| Known strengths | All-round features, large ecosystem, image and voice tools | Writing quality, long documents, coding and agents, safety focus | Multimodal input, Google app integration, long context |
| Training emphasis | RLHF and large-scale product feedback | Constitutional AI and safety research | Multimodal training on Google infrastructure (TPUs) |
| Ecosystem | ChatGPT apps, OpenAI API, Microsoft partnership | Claude apps, Claude API, major cloud platforms | Google Search, Workspace, Android, Google Cloud |
| Free tier | Yes, with limits | Yes, with limits | Yes, with limits |
| Developer access | OpenAI API | Claude API | Gemini API |
How do they really differ? ▶ Watch
- Training data and methods - each company uses its own data mix, fine-tuning examples and alignment techniques, which gives each assistant a distinct "personality", writing style and set of strengths.
- Safety philosophy - all three invest in safety; Anthropic places it at the centre of its mission, which influences how Claude handles sensitive requests.
- Context window - all can process long inputs; the exact limits differ by model and change often.
- Tools and features - image generation, voice mode, web browsing, file creation, memory and agents vary by product and plan.
- Integration - Gemini is strongest inside Google's world; ChatGPT has the broadest consumer feature set and Microsoft ties; Claude is popular with developers and businesses handling long, complex work.
- Pricing - each offers free and paid consumer plans, plus pay-per-token APIs for developers.
Other important AI models and assistants ▶ Watch
| Name | Maker | Notable for |
|---|---|---|
| Llama | Meta | Popular open-weight model family |
| Mistral models | Mistral AI (France) | Efficient open-weight and commercial models |
| DeepSeek | DeepSeek (China) | Low-cost open-weight reasoning models |
| Grok | xAI | Assistant integrated with the X platform |
| Microsoft Copilot | Microsoft | AI built into Windows and Office apps |
| Perplexity | Perplexity AI | AI-powered search with cited sources |
Which one should you use? ▶ Watch
| Your task | What to look for |
|---|---|
| Everyday questions and learning | Any of the three works well; try each free tier |
| Long reports, contracts, books | Large context window and careful reading of long documents |
| Programming and building apps | Strong coding models and agentic coding tools |
| Working inside Gmail, Docs, Android | Gemini's built-in integration |
| Images and voice in one app | Assistants with built-in image generation and voice mode |
| Private or offline use | Open-weight models you can run yourself |
Chapter summary
- ChatGPT (OpenAI), Claude (Anthropic) and Gemini (Google) are all built on large Transformer foundation models.
- They differ in training methods, safety approach, features, integrations and pricing.
- Other major players include Meta Llama, Mistral, DeepSeek, Grok, Microsoft Copilot and Perplexity.
- Test assistants on your own tasks; versions and features change quickly.