Short answer
Give the main session to Opus 5.5 (the default), the hardest reasoning and long-horizon work to Fable 5.1, subagents and bulk edits to Sonnet 5.5 (half the price of Opus 5.5), and light tasks to Haiku 5.5 (1/40 the price of Opus 5.5). Switch any time with /model.
OpusVIP offers four models: Opus 5.5 is the flagship workhorse, Fable 5.1 is Anthropic's most capable model, and Sonnet 5.5 and Haiku 5.5 are strong partners inside Claude Code. Prices are fixed, so start with the numbers, then look at how to split the work.
Price comparison
OpusVIP bills in credits: 1 credit = $0.01 of usage at list price, and 100 credits cost $0.15, so you pay 15% of list price. The credit figures below are exactly what the console log deducts.
| Per 1M tokens | Opus 5.5 | Fable 5.1 | Sonnet 5.5 | Haiku 5.5 |
|---|---|---|---|---|
| List price input / output | $4 / $20 | $10 / $50 | $2 / $10 | $0.10 / $0.50 |
| OpusVIP input / output (credits) | 400 / 2,000 | 1,000 / 5,000 | 200 / 1,000 | 10 / 50 |
| You pay input / output | $0.60 / $3.00 | $1.50 / $7.50 | $0.30 / $1.50 | $0.015 / $0.075 |
| Cache read (credits) | 20 | 25 | 10 | 1 |
| Cache write, 5 min (credits) | 500 | 1,250 | 250 | 12.5 |
| Context | 1M | 1M | 1M | 1M |
Haiku 5.5 prompts above 100K tokens use Anthropic's long-context rates (credits per 1M tokens): input 50, output 250, cache read 5, cache write (5 min) 62.5.
As an illustration, take a task with 1M regular input tokens and 200k output tokens (hypothetical numbers): Opus 5.5 comes to about 800 credits (you pay $1.20), Fable 5.1 about 2,000 ($3.00), Sonnet 5.5 about 400 ($0.60), and Haiku 5.5 about 20 ($0.03).
Who does what in Claude Code
- Opus 5.5: the default for your main session. Complex features, cross-file refactors, architecture and stubborn bugs belong with the strongest everyday model.
- Fable 5.1: Anthropic's most capable model, built for the hardest reasoning and long-running agentic work. Switch with
/model fablewhen you need maximum capability. - Sonnet 5.5: half the price of Opus 5.5, ideal for subagents, bulk mechanical edits and everyday development.
- Haiku 5.5: fast and the lowest price per token, perfect for light tasks, session titles and quick questions.
Claude Code's default aliases map straight to these four models, with no extra configuration.
How to switch
Type /model opus, /model fable, /model sonnet or /model haiku in a session, or start with claude --model sonnet. See Choose and switch models for details.
Test with your own tasks
- Pick three to five tasks you know well and can judge clearly.
- Run each with
/model opusand/model sonnet. - Compare result quality, the number of iterations, and the actual credits spent in the console log.
For model capabilities and release notes, see Anthropic's documentation. All models are listed on the models overview.
Questions? Email hello@opusvip.dev.