Anthropic, the second largest AI vendor after OpenAI, has a powerful family of generative AI models called Claude. These models can perform a variety of tasks, from captioning images and writing emails to solving math and coding challenges.
Because Anthropic's model ecosystem is growing rapidly, it can be difficult to keep track of which Claude models do what. To help you with this, we've created Claude's guide. We will continue to update as new models and upgrades arrive.
claude model
Claude's models are named after literary works such as Haiku, Sonnet, and Opus. The latest ones are:
Claude 3.5 Haiku is a lightweight model. Claude 3.5 Sonet, mid-range model. Claude 3 Opus is Anthropic's flagship model.
It's counterintuitive, but the Claude 3.5 Sonnet (the “midrange” model) is the most capable Claude model to date. This will almost certainly change when the next version of Opus, Claude 3.5 Opus, is released.
All Claude models can analyze not only the text contained in documents, but also images, charts, graphs, and technical diagrams. Featuring a standard 200,000 token context window, the model can also follow multi-step instructions, use tools (such as a stock ticker tracker), and generate structured output in formats such as JSON.
A context window is the amount of data that a model like Claude can analyze before producing new data, whereas tokens are bits that subdivide the raw data (e.g., the syllables “fan”, “tas” in the word “fantastic”) , “tick”, etc.). . 200,000 tokens is equivalent to approximately 150,000 words, or a 600-page novel.
Unlike many leading generative AI models, Anthropic doesn't have access to the internet, so it's not particularly good at answering current events questions. It also cannot generate images, only simple line diagrams.
The main differences between the Claude models are that the Claude 3.5 Sonnet is faster and understands subtle complex instructions better than the Claude 3 Opus. Haiku struggles with sophisticated prompts, but is the quickest of the three models.
Claude model price
Claude models are available through Anthropic's API and managed platforms such as Amazon Bedrock and Google Cloud's Vertex AI.
Anthropic API pricing is as follows:
Claude 3.5 Haiku costs 25 cents per million input tokens (approximately 750,000 words) or $1.25 per million output tokens. Claude 3.5 Sonnet costs $3 per million input tokens or $15 per million output tokens. Claude 3 Opus costs $15 per million input tokens, or $75. per million output tokens
Anthropic provides rapid caching and batch processing to further save execution time.
Prompt caching allows developers to save a specific “prompt context” that can be reused across API calls to a model while batching asynchronous groups of lower priority (and therefore lower cost) model inference requests. I will.
Claude's plans and apps
For individual users and businesses who want to easily interact with Claude models through apps for the web, Android, and iOS, Anthropic offers a free Claude plan with rate limits and other usage restrictions.
Upgrading to one of your company's subscriptions removes these limitations and unlocks new features. The current plan is:
Claude Pro costs $20 per month and comes with 5x rate limits, priority access, and previews of upcoming features.
Business-focused Team ($30 per user per month) adds dashboards to control billing and user management, as well as integrations with codebases and data repositories such as customer relationship management platforms (such as Salesforce) will be done. Toggle enables or disables citations to verify AI-generated claims. (Like other models, Claude sometimes hallucinates.)
Projects are available to both Pro and Team subscribers. This is the ability to base Claude's output on your knowledge base, allowing you to create style guides, interview transcripts, and more. These customers, along with free tier users, utilize Artifacts, a workspace where users can edit and add content such as code, apps, website designs, and other documents generated by Claude. You can also.
For customers who want more information, there's Claude Enterprise, which allows businesses to upload their own data to Claude so Claude can analyze the information and answer questions about it. Claude Enterprise also comes with a larger context window (500,000 tokens), GitHub integration for engineering teams to sync GitHub repositories with Claude, and projects and artifacts.
Precautions
As with all generative AI models, using Claude comes with risks.
Models tend to hallucinate, so they sometimes make mistakes when summarizing or answering questions. It is also trained on public web data, some of which may be copyrighted or subject to restrictive licenses. Anthropic and many other AI vendors argue that the fair use doctrine protects them from claims of copyright infringement. But that hasn't stopped data owners from filing lawsuits.
Anthropic offers policies that protect certain customers from legal battles arising from fair use challenges. However, it does not resolve the ethical issue of using models trained on data without permission.
TechCrunch has a newsletter focused on AI. Sign up here to get it delivered to your inbox every Wednesday.