What Is Claude? The Model Family and Its Coding Tools
Claude is an Anthropic model family and is also the name of its chat product. This article distinguishes the models, the Claude app and Claude Code, including tool execution and model ID boundaries.
Meet this model
Claude is a family of generative models developed by Anthropic for tasks such as text analysis and coding, with image understanding available in vision-capable versions. “Using Claude” may mean the chat app, a model API call or Claude Code. The model generates responses and tool requests; the application supplies files and an execution environment. These layers under the same brand determine how much of a task can actually be completed.
How models and Claude Code divide the work
A Claude model can analyze a project description and generate code. Claude Code is a product organized around coding tasks and can use development tools with the appropriate permissions. A model that writes a suggested fix and a tool that reads the current repository and runs tests provide different functionality.
API developers must also distinguish client tools from server tools. Your application normally executes a custom function and returns its output through tool_result. Official server tools run on Anthropic infrastructure. A tool name appearing in a model response alone does not prove that a file was edited or a web page was read.
Different model lines serve different workloads
On October 20, 2026, the catalog lists Fable 5.1 alongside Opus, Sonnet and Haiku 5.5. The provider positions Fable for demanding reasoning and long-running work, and Haiku for latency-sensitive calls such as classification and extraction. These roles help narrow the candidates but do not replace comparisons on your own tasks.
For a contract spanning dozens of pages, check exceptions and the source passages supporting the answer, not just the fluency of the summary. For a front-end change, inspect the actual layout in a browser as well. This article does not report hands-on Claude testing. The examples describe what to compare, rather than treating documented features as verified delivery results.
Model naming and service-use limits
Starting with the Claude 4.6 generation, canonical model IDs omit dates but identify pinned snapshots. The documentation states that they are not evergreen aliases; some shorter names for earlier models are convenience aliases. Recording only an incomplete name such as claude-sonnet loses important version information and makes upgrade differences harder to explain.
A pinned snapshot does not freeze the surrounding service. Infrastructure updates, including routing or sampling changes, can still affect observable behavior. This article covers Claude hosted interfaces under the applicable service terms. API examples are not an open-weight license. For local deployment, choose a model with explicitly published weights and a license.
Checked sources
These sources support the concepts and methods described here. Examples and test scope are labeled separately; check the version of the tools you use.