Skip to content

Open-source design agent: on-brand stills, decks, diagrams, charts and motion graphics from a chat thread, built on Deep Agents.

2.9 KB
Raw Blame
# Model prices for the cost meter (src/design_agent/harness/meter.py), in USD per million tokens.
#
# Every figure is an estimate, copied by hand from the provider's public pricing page on `date`.
# Providers change prices, discount batch and regional traffic, and bill some features on top; their
# invoice is the source of truth. Re-check the page and update `date` when you change a figure.
#
# Fields per model:
#   input           uncached prompt tokens
#   output          generated tokens
#   cache_write_5m  prompt tokens written to the cache with a 5-minute lifetime
#   cache_write_1h  ... with a 1-hour lifetime (a provider with one write price sets both the same)
#   cache_read      prompt tokens read from the cache
# A `long_context` table holds the prices a request pays once its prompt (cache reads and writes
# included) passes `above_input_tokens`. `date` and `source` may also be set on one model.
# A dated snapshot id (`claude-opus-5-5-20261001`) and a `provider:model` id match their model here.

[anthropic]
source = "https://platform.claude.com/docs/en/about-claude/pricing"
date = 2026-10-17

[anthropic.models.claude-fable-5-1]
input = 10.00
output = 50.00
cache_write_5m = 12.50
cache_write_1h = 20.00
cache_read = 0.25

[anthropic.models.claude-opus-5-5]
input = 4.00
output = 20.00
cache_write_5m = 5.00
cache_write_1h = 8.00
cache_read = 0.20

[anthropic.models.claude-opus-5]
input = 5.00
output = 25.00
cache_write_5m = 6.25
cache_write_1h = 10.00
cache_read = 0.50

[anthropic.models.claude-sonnet-5-5]
input = 2.00
output = 10.00
cache_write_5m = 2.50
cache_write_1h = 4.00
cache_read = 0.10

[anthropic.models.claude-haiku-5-5]
input = 0.10
output = 0.50
cache_write_5m = 0.125
cache_write_1h = 0.20
cache_read = 0.01

[anthropic.models.claude-haiku-5-5.long_context]
above_input_tokens = 100_000
input = 0.50
output = 2.50
cache_write_5m = 0.625
cache_write_1h = 1.00
cache_read = 0.05

[openai]
source = "https://developers.openai.com/api/docs/pricing"
date = 2026-10-17

# Standard tier. OpenAI lists one cache-write price; long context starts above 272K input tokens.
[openai.models."gpt-6.1-sol"]
input = 2.00
output = 10.00
cache_write_5m = 2.50
cache_write_1h = 2.50
cache_read = 0.10

[openai.models."gpt-6.1-sol".long_context]
above_input_tokens = 272_000
input = 4.00
output = 15.00
cache_write_5m = 5.00
cache_write_1h = 5.00
cache_read = 0.20

[openai.models."gpt-6-sol"]
input = 2.00
output = 10.00
cache_write_5m = 2.50
cache_write_1h = 2.50
cache_read = 0.20

[openai.models."gpt-6-sol".long_context]
above_input_tokens = 272_000
input = 4.00
output = 15.00
cache_write_5m = 5.00
cache_write_1h = 5.00
cache_read = 0.40

[openai.models."gpt-5.6-sol"]
input = 4.00
output = 20.00
cache_write_5m = 5.00
cache_write_1h = 5.00
cache_read = 0.40

[openai.models."gpt-5.6-sol".long_context]
above_input_tokens = 272_000
input = 8.00
output = 30.00
cache_write_5m = 10.00
cache_write_1h = 10.00
cache_read = 0.80