The Julius model guide

Anthropic Claude models compared

Compare Claude Sonnet, Opus, Fable, and Haiku: features, context windows, API pricing, and Julius availability.

Try Julius for free

Compare the models

Start free with Julius Lite. Other model access depends on your plan.

Cleaned six-order worksheet and monthly chart showing net sales growing from $1,300 to $1,600 after refunds and deduplication. — Cleaned sales worksheetIllustrative output · Fictional dataView full size

Compare Anthropic models

Sonnet 5.5 emphasizes speed and cost, Opus 5.5 handles sustained coding and knowledge work, and Fable 5.1 targets especially demanding reasoning. All three are choices in Julius. Haiku 5.5 is included here for research and is not available in Julius.

4 results

API prices are USD per million input / output tokens. Julius uses separate plans and credits. See each model for pricing tiers and conditions. Scroll the table to compare all columns.

AI model capabilities, Julius plans, and provider API prices
Model	Consider it for	Published context	Native inputs	Julius access	API input / output
Claude Sonnet 5.5 Anthropic	Everyday work with a balance of speed and cost	1,000,000 tokens	Text and images	Pro, Max, Ultra, and Business	$2 / $10 Anthropic
Claude Opus 5.5 Anthropic	Sustained research, coding, and knowledge work	1,000,000 tokens	Text and images	Pro, Max, Ultra, and Business	$4 / $20 Anthropic
Claude Fable 5.1 Anthropic	Difficult reasoning and complex, multi-step work	1,000,000 tokens	Text and images	Pro, Max, Ultra, and Business	$10 / $50 Anthropic
Claude Haiku 5.5 Anthropic	Researching fast classification and extraction models	1,000,000 tokens	Text and images	Not available	From $0.10 / From $0.50 Anthropic

Explore what you can make with Julius: analyze Excel spreadsheets, create presentations, or build websites and interactive apps. Each model page explains its availability on Julius.

Which Claude model should you choose?

Anthropic recommends Opus for most workloads and positions Sonnet around speed and intelligence. Use the task and your tolerance for latency to narrow the choice.

Everyday analysis with quick iteration

Consider Sonnet for focused file work and frequent follow-up questions. Its sales-cleanup example gives you a small task with an answer you can verify.

Claude Sonnet 5.5: details and uses

Sustained coding and knowledge work

Consider Opus for work that needs sustained reasoning. The conversion example shows why the overall rate and the underlying segments need separate checks.

Claude Opus 5.5: details and uses

Demanding reasoning with several constraints

Consider Fable when the task calls for more intensive reasoning. Anthropic lists higher latency and API prices; test whether the difference helps your actual task.

Claude Fable 5.1: details and uses