Skip to main content
Claude is Anthropic’s language model family, and consistently the strongest option in this catalog for long-form coherence: holding register, terminology, and voice steady across an entire document rather than segment by segment. Those qualities show up in human review rather than in automatic scores, which is why Claude tends to reduce edit distance more than its benchmark position predicts. Every current Claude model carries a 1M-token context window with 128K maximum output and adaptive thinking, and supports text and image input. On Anthropic’s current tokenizer, 1M tokens is roughly 555,000 words — enough to hold most book-length source material in a single pass.
Available in Ollang Workflows — all Claude tiers serve Document Translation, Subtitle Translation, and AI Dubbing. See the Translation catalog.

​
Models

​
Choosing a tier

Claude 5 Fable is Anthropic’s highest-capability model, described as next-generation intelligence for long-running agents. Its thinking is adaptive and always on, and its comparative latency is the slowest in the lineup. Reserve it for genuinely hard source material — dense technical or legal text, structurally ambiguous documents, or work that would otherwise need a second human pass. Claude 5 Opus offers the same 1M context and 128K output, and carries the most recent knowledge cutoff in the family. For quality-critical localization it is the sensible default within Claude. Claude 5 Sonnet is positioned by Anthropic as the best combination of speed and intelligence, and is the throughput-oriented tier of the current family. If a Workflow currently runs Claude 4 or 4.5 Sonnet, this is a straight upgrade. Claude 4.7 Opus, 4.5 Opus, 4.5 Sonnet, and 4 Sonnet remain selectable for Workflows with an existing benchmark. Each is now matched or beaten by its Claude 5 equivalent.

​
Where it fits in a workflow

Claude is the family to reach for when a translation has to read as though one person wrote it: books, courses, help centers, brand-critical marketing, and any document set where terminology drift across sections is the failure mode you are trying to prevent. The 1M-token window means an entire set can be translated in one context rather than chunked, which is what actually produces the consistency. Pair it with Memory and custom instructions — Claude follows detailed style and terminology guidance reliably, so the steering investment compounds across Orders.

​
Trade-offs

  • Fable’s latency is the slowest in the lineup. Use it deliberately on hard material, not as a default.
  • The 4.x tiers are superseded on capability by their Claude 5 equivalents.

​
Reference