Anthropic, the second largest AI provider after OpenAI, has a robust family of generative AI models called Claude. These models can perform a spread of tasks, from captioning images to writing emails to solving math and programming challenges.
Because Anthropic's model ecosystem is growing so quickly, it will possibly be difficult to maintain track of which Claude models are doing what. To enable you, we've put together a guide to Claude, which we'll proceed to update as recent models and upgrades come onto the market.
Claude models
Claude models are named after literary artistic endeavors: Haiku, Sonnet and Opus. The latest are:
- Claude 3 Haikua lightweight model.
- Claude 3.5 sonneta mid-range model.
- Complete 3 papersAnthropic's flagship model.
Counter-intuitively, Claude 3.5 Sonnet – the “mid-range” model – is currently essentially the most powerful Claude model. That will almost definitely change once the following version of Opus, Claude 3.5 Opus, is released.
All Claude models can analyze text in addition to images, charts, graphs and technical diagrams from documents. The models – which have a regular 200,000 token context window – may follow multi-step instructions. Use tools (e.g. stock ticker trackers) and produce structured output in formats corresponding to JSON.
A context window is the quantity of information that a model like Claude can analyze before generating recent data, while tokens are subdivided bits of raw data (just like the syllables “fan,” “tas,” and “tic” within the word “incredible”). . Two hundred thousand tokens are such as about 150,000 words or a 600-page novel.
Unlike many large generative AI models, Anthropics can't access the web, meaning they're not particularly good at answering questions on current events. You can also't create images, only easy line charts.
As for the major differences between the Claude models, Claude 3.5 Sonnet is quicker than Claude 3 Opus and higher understands nuanced and sophisticated instructions. Haiku struggles with demanding prompts, but is the fastest of the three models.
Prices for Claude models
The Claude models can be found through Anthropic's API and managed platforms corresponding to Amazon Bedrock and Google Cloud's Vertex AI.
Here are the costs for the Anthropic API:
- Claude 3 Haiku costs 25 cents per million input tokens (~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
- Complete 3 papers costs $15 per million input tokens or $75 per million output tokens
Anthropic offers fast caching and batching for extra runtime savings.
Prompt caching allows developers to store specific “prompt contexts” that will be reused across API calls to a model, while batching processes asynchronous groups of model inference requests at a lower priority (and due to this fact cheaper cost).
Claude plans and apps
For individual users and businesses who wish to easily interact with Claude models through apps for the net, Android and iOS, Anthropic offers a free Claude plan with tariff caps and other usage restrictions.
Upgrading to one in all the corporate's subscriptions removes these limitations and unlocks recent features. The current plans are:
Claude Pro, which costs $20 per thirty days, offers five times higher plan limits, priority access, and previews of upcoming features.
Team, which is business-focused and costs $30 per user per thirty days, adds a dashboard to regulate billing and user management, in addition to integrations with data repositories corresponding to codebases and customer relationship management platforms (e.g. Salesforce). A switch enables or disables citations to confirm AI-generated claims. (Like all models, Claude hallucinates infrequently.)
Both Pro and Team subscribers get Projects, a feature that anchors Claude's findings into knowledge bases, which will be style guides, interview transcripts, etc. These customers, together with Free Tier users, may access Artifacts, a workspace where users can edit and add content corresponding to code, apps, website designs, and other documents generated by Claude.
For customers who need much more, there may be Claude Enterprise, which allows firms to upload proprietary data into Claude in order that Claude can analyze the data and answer questions on it. Claude Enterprise also has a bigger context window (500,000 tokens), a GitHub integration for engineering teams to sync their GitHub repositories with Claude, and projects and artifacts.
A word of caution
As with all generative AI models, there are risks related to using Claude.
The models occasionally Make mistakes when summarizing or Answering questions due to their tendency to hallucinate. They are also trained on public web data, a few of which could also be copyrighted or subject to a restrictive license. Anthropic and lots of other AI providers argue that the fair use The doctrine protects them from copyright claims. But that hasn’t stopped data owners out of file lawsuits.
Anthropocene offers policies to guard certain customers from litigation related to fair use challenges. However, they don’t resolve the moral dilemma related to using models trained on data without permission.
