Skip to content
Get API Key

Menu

Compare models & routing

Compare models androuting strategies.

See where phantom-1 sends a request, and how that compares with calling one model yourself. Compare prices, context and use cases, then pick your approach.

Get API KeyView Docs
  • Live catalog prices
  • Any two models
  • One API either way
phantom-1
EconomyNemotron 3.5 Lightning FreeLowest price
StandardDeepSeek V4.1 FlashMost coding steps
FrontierClaude Sonnet 5.5The strongest models

Routes each request to a tierSimple tasks don't pay frontier prices.

Side-by-side comparison

Find the rightapproach for your needs.

Compare phantom-1 with calling a model directly on price, use cases, control and setup.

  • Live catalogRefreshed every hour

  • Today's modelsWhat phantom-1 runs now

  • Nothing guessedUnmeasured rows say so

Featurephantom-1(Routing)DefaultEconomy(Direct)Standard(Direct)Frontier(Direct)Manual selection(Direct)
Overall qualitySee Benchmarks
Not measured
Not measured
Not measured
Not measured
VariesDepends on the model
Priceper 1M tokens, in / out
The picked model's priceNo routing fee
FreeNemotron 3.5 Lightning Free
$0.079 / $0.315DeepSeek V4.1 Flash
$2.10 / $10.50Claude Sonnet 5.5
VariesThe model you pick
Latency
Not measured
Not measured
Not measured
Not measured
VariesDepends on the model
Use cases
General purposePhantom picks per request
Simple tasksSummaries, extraction, formatting
Everyday tasksChat, writing, most coding
Hard tasksReasoning, proofs, concurrency bugs
Specialized use casesWhen you know which model to use
Controllability
Your lineupYou pick each tier's model
FullYou name the model
FullYou name the model
FullYou name the model
FullYour own selection logic
Setup
Very lowSend phantom-1 as the model
LowName the model
LowName the model
LowName the model
MediumYou write the selection logic

Model A vs Model B

Compare any two models.

Compare pricing, context limits, capabilities, and data-handling options side by side. Choose the model that fits your workload and budget.

Catalog details reflect the currently listed models. Performance depends on your task; see Benchmarks for measured results.

Each phantom-1 tier has a short list of models. If a model you compare is on a tier’s list, you can put it in your key’s lineup.

Compare the models behind phantom-1

How phantom-1 routes between models

Three models.Your picks.

Pick the model for each tier in your key's lineup. phantom-1 sends each request to one of them.

Learn how routing works

Economy modelLowest price

  • 1M context
  • Free
  • For simple tasks

Standard modelMost coding steps

  • 1M context
  • $0.079 / $0.315 per 1M
  • For everyday tasks

Frontier modelThe strongest models

  • 1M context
  • $2.10 / $10.50 per 1M
  • For the hardest tasks

phantom-1Picks a tier for each new conversation, then keeps it on that model.

What it looks at

  • What the first message asks
  • Tool use and images
  • Prompt length
  • Your lineup and privacy controls
  • Same conversation, same model

Example requests

See how differentrequests get routed.

Where phantom-1 sends each request, and why. phantom-1 made these picks itself.

Shows where it would go. No model answers it, and nothing is stored.

  1. Summarize text

    “Summarize this 10-page research paper.”

    Economy

    Nemotron 3.5 Lightning Free

    Routine text work

  2. Classify content

    “Classify these customer support tickets.”

    Economy

    Nemotron 3.5 Lightning Free

    Routine text work

  3. Write a blog post

    “Write a blog post about AI in healthcare.”

    Standard

    DeepSeek V4.1 Flash

    Everyday task

  4. Analyze data

    “Analyze this dataset and find key insights.”

    Standard

    DeepSeek V4.1 Flash

    Everyday task

  5. Fix a race condition

    “Find the race condition in this worker pool.”

    Frontier

    Claude Sonnet 5.5

    Concurrency bug

  6. Your custom request

    “Generate a marketing image prompt for a climate app.”

    Standard

    DeepSeek V4.1 Flash

    Everyday task

When to use which

Choose the rightapproach.

Both phantom-1 and calling a model directly have their place. Here's how to decide.

Use phantom-1 when…

  • You'd rather not pick a model for each task
  • You want simple tasks to cost less than hard ones
  • You're building a general-purpose app
  • You want one API with no model selection logic
  • You'd rather work on your app than on model choice

Use direct models when…

  • You need one model's specific behavior
  • You need full control over the model and its parameters
  • You're working on a specialized use case
  • You want to try new models as they come out
  • Your workflow is already tuned to one model

Ready to build

Start building with phantom-1.

Send phantom-1 as the model. Phantom picks the tier for each request.

Get API KeyView Docs