Skip to content

Anthropic

Frontier models with a safety-first identity

Models

3

Founded

2021

Country

United States

San Francisco

Anthropic has been building models out of San Francisco, United States, since 2021. Webparam tracks three of its models, spanning reasoning, coding and language: free to call and reaching 1M.

Models

The Anthropic catalogue

Ordered by weekly usage on Webparam.

  • Claude Opus 4.5 by Anthropic
    AnthropicOn request

    Claude Opus 4.5

    anthropic/claude-opus-4.5

    Anthropic's deepest model: frontier reasoning with the judgment for genuinely hard, open-ended work.

    ReasoningTool callingJSON modeStreaming

    200K ctx · Free

  • Claude Sonnet 4.5 by Anthropic
    AnthropicOn request

    Claude Sonnet 4.5

    anthropic/claude-sonnet-4.5

    The default serious coding model, million-token context, dependable agent loops, production temperament.

    CodingTool callingJSON modeStreaming

    1M ctx · Free

  • Claude Haiku 4.5 by Anthropic
    AnthropicOn request

    Claude Haiku 4.5

    anthropic/claude-haiku-4.5

    Near-Sonnet quality at a third of the price, the fast tier of the Claude family.

    LanguageTool callingJSON modeStreaming

    200K ctx · Free

Recent releases

What shipped lately

    Founded by former OpenAI researchers, Anthropic pairs frontier capability with an explicit safety research agenda. Its Claude family (Opus, Sonnet, Haiku) became the default choice for agentic coding and long-document work, with Sonnet 4.5's million-token context a favourite for repository-scale tasks.

    Strengths

    • Best-in-class agentic coding (Sonnet line)
    • Long-context reliability at 1M tokens
    • Safety research that shapes the industry

    Use Anthropic models through Webparam

    Every model above is callable through one OpenAI-compatible convention: change the model ID, keep the request.

    Chat completions reference →Claude Opus 4.5 model details →

    Related providers

    Labs in the same conversation

    All providers →