Matching models 4 Rows visible after provider and capability filters.
Provider scope ANTHROPIC Current provider slice for this table view.
Pricing flags Any cache ยท Any batch Use these to narrow the list to operationally usable models.
Sort logic Latest first Choose the cost signal that matches the workload you are pricing.
Apply

Newest tracked models appear first by default. Switch to a price sort when the workload is stable enough to optimize for cost instead of recency.

Model Provider Input / 1M Output / 1M Blended / 1M Capabilities Updated
Claude Sonnet 4.6
claude-sonnet-4.6
Anthropic Docs
Anthropic $3.00 $15.00 $6.00
Cached $0.300 No batch price multimodal
Context 200000
Updated Mar 30, 02:54
Primary freshness signal
Claude Opus 4.6
claude-opus-4.6
Anthropic Docs
Anthropic $5.00 $25.00 $10.00
Cached $0.500 No batch price multimodal
Context 200000
Updated Mar 30, 02:54
Primary freshness signal
Claude Haiku 4.5
claude-haiku-4.5
Anthropic Docs
Anthropic $1.00 $5.00 $2.00
Cached $0.100 No batch price multimodal
Context 200000
Updated Mar 30, 02:54
Primary freshness signal
Claude Haiku 3.5
claude-haiku-3.5
Anthropic Docs
Anthropic $0.800 $4.00 $1.60
Cached $0.080 No batch price text
Context 200000
Updated Mar 30, 02:54
Primary freshness signal

When to use the list

The list is best for scanning the market, checking freshness, and eliminating clearly expensive options before you do workload-specific math.

When to open compare

Use compare after the shortlist is small and the workload is fixed. That gives a cleaner decision than sorting one price field in isolation.

When to open detail

Open the model detail page when you need source links, pricing history, context limits, or a quick reference workload estimate before committing to a choice.