Become a 10x Brand Engineer with Cuppa AI See Who's Hiring ↗
Getting StartedAnthropic API

Anthropic API

Anthropic (Claude) API Setup

Connect your Anthropic API key to use Claude models for content generation. Claude Opus 5.5 is the current daily default — Fable-level work at 4/4 / 20. Claude Sonnet 5.5 is the new volume pick at 2/2 / 10. Same Anthropic key you already use, no extra setup.

Why Claude?

Anthropic’s Claude models are known for:

  • Natural writing style - less “AI-sounding”
  • Long-form content - handles complex, lengthy articles well
  • Instruction following - excellent at matching tone and style
  • Thoughtful responses - better at nuanced topics

Many Cuppa users prefer Claude for their final, published content.

Available Models

ModelIDBest ForInput / Output (per 1M tokens)Notes
Claude Opus 5.5claude-opus-5-5Daily default. Fable-level work, clearer writing4/4 / 20Marked NEW. 1M context, 128K output. Cache reads $0.20. Typical jobs ~40% cheaper than Opus 5
Claude Fable 5.1claude-fable-5-1Flagship public. Long-horizon writing, research, agents10/10 / 50NEW. 1M context, 128K output. Cache reads $0.25
Claude Mythos 5.1claude-mythos-5-1Same as Fable 5.110/10 / 50NEW. Anthropic trusted access only (Glasswing / CVP / LSVP)
Claude Sonnet 5.5claude-sonnet-5-5Everyday + agentic at volume. Near Opus 5.5 on knowledge work, clearer writing2/2 / 10Marked NEW. 1M context, 128K output. Fastest Sonnet yet, cheaper than Sonnet 5
Claude Opus 5claude-opus-5Previous premium5/5 / 25Stays in the picker
Claude Sonnet 5claude-sonnet-5Previous Sonnet3/3 / 15Stays in the picker
Claude Fable 5claude-fable-5Previous Fable10/10 / 50Stays in the picker
Claude Haiku 4.5claude-haiku-4-5Bulk drafts1/1 / 5Cheapest Claude

Opus 5, Fable 5, Sonnet 5, Sonnet 4.5, and Opus 4.8 stay in the picker. Switch back in one click if 5.5 is not the winner on your content.

Claude Mythos 5.1 is the same weights as Fable 5.1 with looser cyber and life-sciences safeguards. If your Anthropic key does not have trusted access, it will not run. Most content work should stay on Opus 5.5 or Fable 5.1.

Recommendation: Default to Opus 5.5. Use Fable 5.1 for high-stakes long-form and Agent Chat. Use Sonnet 5.5 or Haiku 4.5 for volume.

List the live catalog with GET /v1/meta/models or MCP list_models. Prices verified 2026-09-28 (Anthropic).

Step 1: Create an Anthropic Account

  1. Go to console.anthropic.com
  2. Click Sign Up
  3. Verify your email
  4. Complete account setup

Step 2: Add Payment Method

Anthropic requires a payment method before generating API keys:

  1. Go to Settings → Plans & Billing
  2. Click Add payment method
  3. Enter your credit card details
  4. Add credits ($5-20 to start)

Step 3: Generate Your API Key

  1. Go to console.anthropic.com/settings/keys
  2. Click Create Key
  3. Name it “Cuppa” (optional)
  4. Click Create Key
  5. Copy the key immediately - you won’t see it again!

Your key looks like: sk-ant-api03-xxxxxxxxxxxxxxxxxxxx

Step 4: Add Key to Cuppa

  1. Log in to app.cuppa.ai
  2. Go to Team Settings → API Keys
  3. Find the Anthropic section
  4. Paste your API key
  5. Click Save

If Anthropic is already connected, you do not need a new key for Opus 5.5, Sonnet 5.5, or Fable 5.1.

Using Claude in Cuppa

Once connected, Opus 5.5, Sonnet 5.5, and Fable 5.1 appear in every text model selector (Articles, Agent Chat, Editor, Workbench, Compare Models, Content Planner, Social, Bulk, Programmatic):

  1. Open any generation flow
  2. Click the model selector
  3. Choose Claude Opus 5.5 (marked NEW)
  4. Generate as usual

Via API / CLI / MCP, pass model: "claude-opus-5-5", model: "claude-sonnet-5-5", or model: "claude-fable-5-1".

Single Generation

  1. Go to Create Content → Single Generation
  2. In the model dropdown, select Claude Opus 5.5 (or Fable 5.1 / Sonnet 5.5)
  3. Continue with normal generation

Bulk Generation

  1. Go to Create Content → Bulk Generation
  2. Select a Claude model in settings
  3. All articles in the batch will use Claude

In Presets

  1. Go to Saved Presets → My Presets
  2. Create or edit a preset
  3. Set the default model to Claude
  4. All content using that preset will use Claude

Claude vs GPT Comparison

FeatureClaudeGPT-4o
Writing styleMore naturalMore structured
Long contentExcellentGood
Following instructionsExcellentExcellent
SpeedFastFast
Cost (per article)~$0.10-0.30~$0.10-0.25

When to use Claude:

  • Final, published content
  • Brand voice matching
  • Nuanced topics
  • Long-form guides

When to use GPT:

  • Bulk drafts
  • Technical content
  • Structured formats
  • Cost optimization

Rate Limits

Anthropic has rate limits based on your usage tier:

TierRequirementRequests/min
Tier 1New account60
Tier 2$100+ spent1,000
Tier 3$500+ spent2,000

For bulk generation, Tier 2 is recommended.

Cost Management

Monitoring Usage

  1. Go to console.anthropic.com/settings/usage
  2. View current spend and limits
  3. Set spending limits if needed

Typical Costs

Content TypeModelEstimated Cost
Single article (2000 words)Claude Haiku 4.5$0.02-0.05
Single article (2000 words)Claude Sonnet 5.5$0.05-0.13
Single article (2000 words)Claude Sonnet 5$0.08-0.20
Single article (2000 words)Claude Opus 5.5$0.08-0.25
Single article (2000 words)Claude Opus 5$0.10-0.30
Single article (2000 words)Claude Fable 5.1$0.20-0.60
Bulk project (100 articles)Claude Haiku 4.5$2-5
Bulk project (100 articles)Claude Sonnet 5.5$5-13
Bulk project (100 articles)Claude Sonnet 5$8-20

Important Note

OpenAI is still required even when using Claude. Cuppa uses GPT-4o-mini for certain internal features:

  • Outline generation
  • SEO metadata
  • Content optimization

Make sure you have both keys connected for the best experience.

Troubleshooting

”Invalid API key”

  • Check you copied the full key (starts with sk-ant-)
  • Verify the key exists in Anthropic console
  • Create a new key if needed

”Rate limit exceeded”

  • Wait 1 minute and retry
  • Check your Anthropic tier
  • Consider using GPT for bulk, Claude for final edits

”Insufficient credits”

”Model not available”

  • Ensure you’re using a current model name
  • Some older models may be deprecated
  • Try Claude Opus 5.5, Claude Fable 5.1, or Claude Sonnet 5.5

FAQ

Do I need both OpenAI and Anthropic?

OpenAI is required. Anthropic is optional but recommended for higher-quality content.

Can I use Claude for everything?

You can use Claude for all main content generation. Some internal Cuppa features still require OpenAI.

Do I need a new API key for Opus 5.5, Sonnet 5.5, or Fable 5.1?

No. Same Anthropic key. Open any model selector and pick Claude Opus 5.5, Claude Sonnet 5.5, or Claude Fable 5.1.

Which Claude model is best for articles?

Opus 5.5 for most articles. Fable 5.1 for high-stakes long-form and agent work. Sonnet 5.5 or Haiku 4.5 for volume.

What is Mythos 5.1?

Same model as Fable 5.1 with reduced cyber/bio safeguards. Anthropic only enables it for trusted-access programs. If it is not on your key, use Fable 5.1.

Is Claude better than GPT?

For natural-sounding content, many users prefer Claude. For structured/technical content, GPT often works better. Try both!

How do I switch between models?

Select the model in the generation settings. You can also set a default in your presets.

Related Docs

  • OpenAI API Setup - Required for all users
  • Model Comparison - Detailed model comparison
  • Presets Guide - Save your preferred model settings
OpenAI APIPerplexity API