Claude vs. GPT: Which is better for writing?
In Type, you can summon writing assistance from two different AI providers: OpenAI and Anthropic. OpenAI's GPT models tend to be stronger at complex reasoning tasks, whereas Anthropic's Claude models tend to be more creative and expressive.
Depending on the type of writing you do, you may find one AI better suited than the other.
Here's a short guide for finding the AI that will harmonize with your writing.
Claude is more expressive, but less intelligent
Both Claude models tend to be more creative and expressive than their GPT equivalent. It's almost as if Claude has read every great piece of literature from the last 300 years and every entertaining blog post on the Internet.
Claude also has a much larger context window than any GPT model, which means it can process longer inputs and generate longer outputs. By default, it can process up to 75,000 words or 100,000 tokens in one request. (note: in Type, there are some limits in place to ensure certain requests don't fail)
Here's a more specific breakdown of Claude's relative strengths and weaknesses.
Able to mimic a wide range of writing styles
Particularly good at creative writing tasks (ex. writing dialogue)
Much larger context window
Slightly weaker at complex reasoning tasks
Less able to follow detailed instructions
Less multilingual capabilities
If you're writing content where voice and style are particularly important, Claude will likely give you the best writing support. When it comes to handling more complex requests, though, Claude can sometimes go off the rails – though usually in an interesting way.
GPT is more intelligent, but less expressive
Similar to Anthropic, OpenAI offers two primary options within their GPT family of models, GPT-3.5 Turbo and GPT-4.
Both models tend to be strong at complex reasoning tasks and following instructions, but they're less creative and expressive compared to their equivalent Claude model.
Excellent at complex reasoning tasks
Able to reliably follow complex, multi-step instructions
More factual and less likely to hallucinate
More multilingual capabilities (GPT-4 can generate text in 200+ languages)
Less expressive writing style that can come across as dry or robotic
Smaller context window limits length of inputs and outputs
If you're writing tutorials, how-tos, or something like a memo, you may find GPT particularly useful.
Comparing the different Claude and GPT models
Both the Claude and GPT family of models offer a fast, less-intelligent model and a more intelligent, slower model.
In the case of Anthropic, there are two different Claude models: Claude V2 and Claude Instant. Claude V2 is the most intelligent version of Claude, but it's much slower (and more expensive if you use Anthropic's API directly). And OpenAI's most popular two models are GPT-4 and GPT-3.5 Turbo. GPT-4 is the most intelligent and capable model in the GPT family, whereas GPT tends to be much faster. In reality, OpenAI offers a wider assortment of models but we've excluded most of them from this analysis since they're focused on serving other use cases.
In Type, you can easily toggle between using Claude or GPT models in your settings.
When Power mode is enabled in Type, every AI feature will use the most powerful model from the preferred provider, such as GPT-4 or Claude V2. In Speed mode, Type's AI will use either GPT-3.5 Turbo or Claude Instant.
Which model is better for writing?
As you can see, it depends! Claude's family of models tend to be more expressive and creative – which can be great for creative writing or marketing content. And GPT tends to be better at working with more factual or complex content (think: educational content, how-to's, technical guides, etc.)
If you want to compare them for yourself, you can sign up for Type and toggle between Claude and GPT in your settings. About Type: Type is an AI-powered document editor that helps you write more, better, faster.