From 9621a60d27ceb3aca1fe4ed39bff6559b52b17a1 Mon Sep 17 00:00:00 2001 From: Kerry Lu Date: Mon, 12 Oct 2026 21:29:02 -0700 Subject: [PATCH] test(cost): add long context tier cost cases for gpt-6 (272k) and claude-haiku-5-5 (100k) (#46147) --- .../cost_calculation/cases/anthropic.py | 77 ++++++ .../cost_calculation/cases/azure.py | 251 ++++++++++++++++++ .../cases/bedrock_converse.py | 79 ++++++ .../cost_calculation/cases/misc.py | 30 ++- .../cost_calculation/cases/openai.py | 241 +++++++++++++++++ .../integration/cost_calculation/cost_map.py | 65 ++++- .../cost_calculation/cost_tracking_case.py | 28 ++ 7 files changed, 764 insertions(+), 7 deletions(-) diff --git a/tests/integration/cost_calculation/cases/anthropic.py b/tests/integration/cost_calculation/cases/anthropic.py index cadbca9b96f..ca97898bea9 100644 --- a/tests/integration/cost_calculation/cases/anthropic.py +++ b/tests/integration/cost_calculation/cases/anthropic.py @@ -5,6 +5,7 @@ PARITY pairs the streamed case with its plain twin where both bill the same row. from typing import Final from integration.cost_calculation.cost_tracking_case import ( + with_usage, CostTrackingTestCase, ExactExpected, FailureDetails, @@ -6705,7 +6706,83 @@ CLAUDE_OPUS_5_5_RESPONSES_BRIDGED_JSON: Final = CLAUDE_OPUS_5_5_MESSAGES_INPUT_T } ) +CLAUDE_HAIKU_5_5_TIERED_INPUT_ABOVE_100K: Final = with_usage( + CLAUDE_HAIKU_5_5_INPUT_TEXT, + "claude-haiku-5-5-tiered_input_above_100k", + { + "input_tokens": 150000, + "output_tokens": 480, + "cache_read_input_tokens": 0, + "cache_creation_input_tokens": 0, + "cache_creation": {"ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 0}, + }, + ExactExpected(spend=0.33528, input_cost=0.33, output_cost=0.00528, prompt_tokens=150000, completion_tokens=480), +) + +CLAUDE_HAIKU_5_5_TIERED_AT_100K_BOUNDARY: Final = with_usage( + CLAUDE_HAIKU_5_5_INPUT_TEXT, + "claude-haiku-5-5-tiered_at_100k_boundary", + { + "input_tokens": 100000, + "output_tokens": 480, + "cache_read_input_tokens": 0, + "cache_creation_input_tokens": 0, + "cache_creation": {"ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 0}, + }, + ExactExpected(spend=0.11264, input_cost=0.11, output_cost=0.00264, prompt_tokens=100000, completion_tokens=480), +) + +CLAUDE_HAIKU_5_5_TIERED_CACHE_READ_ABOVE_100K: Final = with_usage( + CLAUDE_HAIKU_5_5_CACHE_READ, + "claude-haiku-5-5-tiered_cache_read_above_100k", + { + "input_tokens": 4096, + "output_tokens": 480, + "cache_read_input_tokens": 110592, + "cache_creation_input_tokens": 0, + "cache_creation": {"ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 0}, + }, + ExactExpected( + spend=0.03862144, input_cost=0.03334144, output_cost=0.00528, prompt_tokens=114688, completion_tokens=480 + ), +) + +CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_5M_ABOVE_100K: Final = with_usage( + CLAUDE_HAIKU_5_5_CACHE_WRITE_5M, + "claude-haiku-5-5-tiered_cache_write_5m_above_100k", + { + "input_tokens": 4096, + "output_tokens": 480, + "cache_read_input_tokens": 0, + "cache_creation_input_tokens": 110592, + "cache_creation": {"ephemeral_5m_input_tokens": 110592, "ephemeral_1h_input_tokens": 0}, + }, + ExactExpected( + spend=0.3184192, input_cost=0.3131392, output_cost=0.00528, prompt_tokens=114688, completion_tokens=480 + ), +) + +CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_1H_ABOVE_100K: Final = with_usage( + CLAUDE_HAIKU_5_5_CACHE_WRITE_1H, + "claude-haiku-5-5-tiered_cache_write_1h_above_100k", + { + "input_tokens": 4096, + "output_tokens": 480, + "cache_read_input_tokens": 0, + "cache_creation_input_tokens": 110592, + "cache_creation": {"ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 110592}, + }, + ExactExpected( + spend=0.500896, input_cost=0.495616, output_cost=0.00528, prompt_tokens=114688, completion_tokens=480 + ), +) + CASES: Final[tuple[CostTrackingTestCase, ...]] = ( + CLAUDE_HAIKU_5_5_TIERED_INPUT_ABOVE_100K, + CLAUDE_HAIKU_5_5_TIERED_AT_100K_BOUNDARY, + CLAUDE_HAIKU_5_5_TIERED_CACHE_READ_ABOVE_100K, + CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_5M_ABOVE_100K, + CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_1H_ABOVE_100K, CLAUDE_SONNET_5_RESPONSES_BRIDGED_JSON, CLAUDE_SONNET_5_5_RESPONSES_BRIDGED_JSON, CLAUDE_HAIKU_5_5_RESPONSES_BRIDGED_JSON, diff --git a/tests/integration/cost_calculation/cases/azure.py b/tests/integration/cost_calculation/cases/azure.py index df96d1d9993..173cf0626c1 100644 --- a/tests/integration/cost_calculation/cases/azure.py +++ b/tests/integration/cost_calculation/cases/azure.py @@ -5,6 +5,7 @@ PARITY pairs the streamed case with its plain twin where both bill the same row. from typing import Final from integration.cost_calculation.cost_tracking_case import ( + with_usage, BinaryResponse, CostTrackingTestCase, Deployment, @@ -5621,11 +5622,261 @@ AZURE_GPT_6_1_SOL_MESSAGES_BRIDGED_JSON: Final = AZURE_GPT_6_1_SOL_INPUT_TEXT.mo } ) +AZURE_GPT_6_SOL_TIERED_INPUT_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_SOL_INPUT_TEXT, + "azure-gpt-6-sol-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=1.389476, input_cost=1.38, output_cost=0.009476, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_SOL_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_SOL_INPUT_TEXT, + "azure-gpt-6-sol-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=0.561476, input_cost=0.552, output_cost=0.009476, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_SOL_TIERED_AT_272K_BOUNDARY: Final = with_usage( + AZURE_GPT_6_SOL_INPUT_TEXT, + "azure-gpt-6-sol-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.630338, input_cost=0.6256, output_cost=0.004738, prompt_tokens=272000, completion_tokens=412), +) + +AZURE_GPT_6_SOL_TIERED_FLEX_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_SOL_SERVICE_TIER_FLEX, + "azure-gpt-6-sol-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.694738, input_cost=0.69, output_cost=0.004738, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_SOL_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_SOL_SERVICE_TIER_PRIORITY, + "azure-gpt-6-sol-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=2.778952, input_cost=2.76, output_cost=0.018952, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_LUNA_TIERED_INPUT_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_LUNA_INPUT_TEXT, + "azure-gpt-6-luna-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected( + spend=0.0785356, input_cost=0.078, output_cost=0.0005356, prompt_tokens=300000, completion_tokens=412 + ), +) + +AZURE_GPT_6_LUNA_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_LUNA_INPUT_TEXT, + "azure-gpt-6-luna-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected( + spend=0.0317356, input_cost=0.0312, output_cost=0.0005356, prompt_tokens=300000, completion_tokens=412 + ), +) + +AZURE_GPT_6_LUNA_TIERED_AT_272K_BOUNDARY: Final = with_usage( + AZURE_GPT_6_LUNA_INPUT_TEXT, + "azure-gpt-6-luna-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected( + spend=0.0356278, input_cost=0.03536, output_cost=0.0002678, prompt_tokens=272000, completion_tokens=412 + ), +) + +AZURE_GPT_6_LUNA_TIERED_FLEX_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_LUNA_SERVICE_TIER_FLEX, + "azure-gpt-6-luna-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected( + spend=0.0392678, input_cost=0.039, output_cost=0.0002678, prompt_tokens=300000, completion_tokens=412 + ), +) + +AZURE_GPT_6_LUNA_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_LUNA_SERVICE_TIER_PRIORITY, + "azure-gpt-6-luna-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected( + spend=0.0785356, input_cost=0.078, output_cost=0.0005356, prompt_tokens=300000, completion_tokens=412 + ), +) + +AZURE_GPT_6_ASTRA_TIERED_INPUT_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_ASTRA_INPUT_TEXT, + "azure-gpt-6-astra-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=7.5515, input_cost=7.5, output_cost=0.0515, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_ASTRA_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_ASTRA_INPUT_TEXT, + "azure-gpt-6-astra-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=3.0515, input_cost=3.0, output_cost=0.0515, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_ASTRA_TIERED_AT_272K_BOUNDARY: Final = with_usage( + AZURE_GPT_6_ASTRA_INPUT_TEXT, + "azure-gpt-6-astra-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=3.42575, input_cost=3.4, output_cost=0.02575, prompt_tokens=272000, completion_tokens=412), +) + +AZURE_GPT_6_ASTRA_TIERED_FLEX_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_ASTRA_SERVICE_TIER_FLEX, + "azure-gpt-6-astra-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=3.77575, input_cost=3.75, output_cost=0.02575, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_ASTRA_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_ASTRA_SERVICE_TIER_PRIORITY, + "azure-gpt-6-astra-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=7.5515, input_cost=7.5, output_cost=0.0515, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_1_SOL_TIERED_INPUT_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_1_SOL_INPUT_TEXT, + "azure-gpt-6.1-sol-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=1.449888, input_cost=1.44, output_cost=0.009888, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_1_SOL_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_1_SOL_INPUT_TEXT, + "azure-gpt-6.1-sol-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=0.585888, input_cost=0.576, output_cost=0.009888, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_1_SOL_TIERED_AT_272K_BOUNDARY: Final = with_usage( + AZURE_GPT_6_1_SOL_INPUT_TEXT, + "azure-gpt-6.1-sol-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.657744, input_cost=0.6528, output_cost=0.004944, prompt_tokens=272000, completion_tokens=412), +) + +AZURE_GPT_6_1_SOL_TIERED_FLEX_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_1_SOL_SERVICE_TIER_FLEX, + "azure-gpt-6.1-sol-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.724944, input_cost=0.72, output_cost=0.004944, prompt_tokens=300000, completion_tokens=412), +) + +AZURE_GPT_6_1_SOL_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + AZURE_GPT_6_1_SOL_SERVICE_TIER_PRIORITY, + "azure-gpt-6.1-sol-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=1.449888, input_cost=1.44, output_cost=0.009888, prompt_tokens=300000, completion_tokens=412), +) + CASES: Final[tuple[CostTrackingTestCase, ...]] = ( AZURE_GPT_6_SOL_MESSAGES_BRIDGED_JSON, AZURE_GPT_6_LUNA_MESSAGES_BRIDGED_JSON, AZURE_GPT_6_ASTRA_MESSAGES_BRIDGED_JSON, AZURE_GPT_6_1_SOL_MESSAGES_BRIDGED_JSON, + AZURE_GPT_6_SOL_TIERED_INPUT_ABOVE_272K, + AZURE_GPT_6_SOL_TIERED_CACHE_READ_ABOVE_272K, + AZURE_GPT_6_SOL_TIERED_AT_272K_BOUNDARY, + AZURE_GPT_6_SOL_TIERED_FLEX_ABOVE_272K, + AZURE_GPT_6_SOL_TIERED_PRIORITY_ABOVE_272K, + AZURE_GPT_6_LUNA_TIERED_INPUT_ABOVE_272K, + AZURE_GPT_6_LUNA_TIERED_CACHE_READ_ABOVE_272K, + AZURE_GPT_6_LUNA_TIERED_AT_272K_BOUNDARY, + AZURE_GPT_6_LUNA_TIERED_FLEX_ABOVE_272K, + AZURE_GPT_6_LUNA_TIERED_PRIORITY_ABOVE_272K, + AZURE_GPT_6_ASTRA_TIERED_INPUT_ABOVE_272K, + AZURE_GPT_6_ASTRA_TIERED_CACHE_READ_ABOVE_272K, + AZURE_GPT_6_ASTRA_TIERED_AT_272K_BOUNDARY, + AZURE_GPT_6_ASTRA_TIERED_FLEX_ABOVE_272K, + AZURE_GPT_6_ASTRA_TIERED_PRIORITY_ABOVE_272K, + AZURE_GPT_6_1_SOL_TIERED_INPUT_ABOVE_272K, + AZURE_GPT_6_1_SOL_TIERED_CACHE_READ_ABOVE_272K, + AZURE_GPT_6_1_SOL_TIERED_AT_272K_BOUNDARY, + AZURE_GPT_6_1_SOL_TIERED_FLEX_ABOVE_272K, + AZURE_GPT_6_1_SOL_TIERED_PRIORITY_ABOVE_272K, AZURE_GPT_5_4_MINI_INPUT_TEXT, AZURE_GPT_5_4_MINI_CACHE_READ, AZURE_GPT_5_4_MINI_AUDIO_INPUT, diff --git a/tests/integration/cost_calculation/cases/bedrock_converse.py b/tests/integration/cost_calculation/cases/bedrock_converse.py index 0b7e7df34a3..45ac7398d73 100644 --- a/tests/integration/cost_calculation/cases/bedrock_converse.py +++ b/tests/integration/cost_calculation/cases/bedrock_converse.py @@ -5,6 +5,7 @@ PARITY pairs the streamed case with its plain twin where both bill the same row. from typing import Final from integration.cost_calculation.cost_tracking_case import ( + with_usage, CostTrackingTestCase, Deployment, EventStreamEvent, @@ -5981,6 +5982,79 @@ US_ANTHROPIC_CLAUDE_FABLE_5_1_RESPONSES_BRIDGED_JSON: Final = US_ANTHROPIC_CLAUD } ) +US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_INPUT_ABOVE_100K: Final = with_usage( + US_ANTHROPIC_CLAUDE_HAIKU_5_5_INPUT_TEXT, + "us.anthropic.claude-haiku-5-5-tiered_input_above_100k", + { + "inputTokens": 150000, + "outputTokens": 480, + "totalTokens": 150480, + "cacheReadInputTokens": 0, + "cacheWriteInputTokens": 0, + }, + ExactExpected(spend=0.36576, input_cost=0.36, output_cost=0.00576, prompt_tokens=150000, completion_tokens=480), +) + +US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_AT_100K_BOUNDARY: Final = with_usage( + US_ANTHROPIC_CLAUDE_HAIKU_5_5_INPUT_TEXT, + "us.anthropic.claude-haiku-5-5-tiered_at_100k_boundary", + { + "inputTokens": 100000, + "outputTokens": 480, + "totalTokens": 100480, + "cacheReadInputTokens": 0, + "cacheWriteInputTokens": 0, + }, + ExactExpected(spend=0.12288, input_cost=0.12, output_cost=0.00288, prompt_tokens=100000, completion_tokens=480), +) + +US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_CACHE_READ_ABOVE_100K: Final = with_usage( + US_ANTHROPIC_CLAUDE_HAIKU_5_5_CACHE_READ, + "us.anthropic.claude-haiku-5-5-tiered_cache_read_above_100k", + { + "inputTokens": 4096, + "outputTokens": 480, + "totalTokens": 115168, + "cacheReadInputTokens": 110592, + "cacheWriteInputTokens": 0, + }, + ExactExpected( + spend=0.04213248, input_cost=0.03637248, output_cost=0.00576, prompt_tokens=114688, completion_tokens=480 + ), +) + +US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_5M_ABOVE_100K: Final = with_usage( + US_ANTHROPIC_CLAUDE_HAIKU_5_5_CACHE_WRITE_5M, + "us.anthropic.claude-haiku-5-5-tiered_cache_write_5m_above_100k", + { + "inputTokens": 4096, + "outputTokens": 480, + "totalTokens": 115168, + "cacheReadInputTokens": 0, + "cacheWriteInputTokens": 110592, + "cacheDetails": [{"ttl": "5m", "inputTokens": 110592}], + }, + ExactExpected( + spend=0.3473664, input_cost=0.3416064, output_cost=0.00576, prompt_tokens=114688, completion_tokens=480 + ), +) + +US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_1H_ABOVE_100K: Final = with_usage( + US_ANTHROPIC_CLAUDE_HAIKU_5_5_CACHE_WRITE_1H, + "us.anthropic.claude-haiku-5-5-tiered_cache_write_1h_above_100k", + { + "inputTokens": 4096, + "outputTokens": 480, + "totalTokens": 115168, + "cacheReadInputTokens": 0, + "cacheWriteInputTokens": 110592, + "cacheDetails": [{"ttl": "1h", "inputTokens": 110592}], + }, + ExactExpected( + spend=0.546432, input_cost=0.540672, output_cost=0.00576, prompt_tokens=114688, completion_tokens=480 + ), +) + CASES: Final[tuple[CostTrackingTestCase, ...]] = ( US_ANTHROPIC_CLAUDE_SONNET_5_RESPONSES_BRIDGED_JSON, US_ANTHROPIC_CLAUDE_SONNET_5_5_RESPONSES_BRIDGED_JSON, @@ -5989,6 +6063,11 @@ CASES: Final[tuple[CostTrackingTestCase, ...]] = ( US_ANTHROPIC_CLAUDE_HAIKU_5_5_RESPONSES_BRIDGED_JSON, US_ANTHROPIC_CLAUDE_FABLE_5_RESPONSES_BRIDGED_JSON, US_ANTHROPIC_CLAUDE_FABLE_5_1_RESPONSES_BRIDGED_JSON, + US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_INPUT_ABOVE_100K, + US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_AT_100K_BOUNDARY, + US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_CACHE_READ_ABOVE_100K, + US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_5M_ABOVE_100K, + US_ANTHROPIC_CLAUDE_HAIKU_5_5_TIERED_CACHE_WRITE_1H_ABOVE_100K, ANTHROPIC_CLAUDE_SONNET_5_V1_0_INPUT_TEXT, ANTHROPIC_CLAUDE_SONNET_5_V1_0_CACHE_READ, ANTHROPIC_CLAUDE_SONNET_5_V1_0_CACHE_WRITE_5M, diff --git a/tests/integration/cost_calculation/cases/misc.py b/tests/integration/cost_calculation/cases/misc.py index 24b57c3e93c..389e552e44c 100644 --- a/tests/integration/cost_calculation/cases/misc.py +++ b/tests/integration/cost_calculation/cases/misc.py @@ -1002,7 +1002,9 @@ AZURE_AI_CLAUDE_SONNET_4_5_DEPLOYMENT_MESSAGES_WEB_SEARCH: Final = CostTrackingT request={ "model": "$MODEL", "max_tokens": 412, - "messages": [{"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_messages_web_search_two_requests"}], + "messages": [ + {"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_messages_web_search_two_requests"} + ], "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 3}], }, response=JsonResponse( @@ -1037,7 +1039,12 @@ AZURE_AI_CLAUDE_SONNET_4_5_DEPLOYMENT_MESSAGES_WEB_SEARCH_STREAM: Final = CostTr request={ "model": "$MODEL", "max_tokens": 412, - "messages": [{"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_messages_web_search_two_requests_stream"}], + "messages": [ + { + "role": "user", + "content": "azure-ai-claude-sonnet-4-5-deployment_rates_messages_web_search_two_requests_stream", + } + ], "stream": True, "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 3}], }, @@ -1090,7 +1097,9 @@ AZURE_AI_CLAUDE_SONNET_4_5_DEPLOYMENT_MESSAGES_WEB_SEARCH_ZERO_RATE: Final = Cos request={ "model": "$MODEL", "max_tokens": 412, - "messages": [{"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_zero_search_rate_messages_two_requests"}], + "messages": [ + {"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_zero_search_rate_messages_two_requests"} + ], "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 3}], }, response=JsonResponse( @@ -1124,7 +1133,9 @@ AZURE_AI_CLAUDE_SONNET_4_5_DEPLOYMENT_MESSAGES_NO_SEARCH: Final = CostTrackingTe request={ "model": "$MODEL", "max_tokens": 412, - "messages": [{"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_messages_no_server_tool_use"}], + "messages": [ + {"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_messages_no_server_tool_use"} + ], "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 3}], }, response=JsonResponse( @@ -1156,7 +1167,9 @@ AZURE_AI_CLAUDE_SONNET_4_5_DEPLOYMENT_CHAT_WEB_SEARCH: Final = CostTrackingTestC deployment=Deployment(input_cost_per_token=1e-06, output_cost_per_token=5e-06, search_cost_per_query=0.01), request={ "model": "$MODEL", - "messages": [{"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_chat_web_search_three_requests"}], + "messages": [ + {"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_chat_web_search_three_requests"} + ], "stream": False, "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 5}], "allowed_openai_params": ["web_search_options"], @@ -1190,7 +1203,12 @@ AZURE_AI_CLAUDE_SONNET_4_5_DEPLOYMENT_CHAT_WEB_SEARCH_STREAM: Final = CostTracki deployment=Deployment(input_cost_per_token=1e-06, output_cost_per_token=5e-06, search_cost_per_query=0.01), request={ "model": "$MODEL", - "messages": [{"role": "user", "content": "azure-ai-claude-sonnet-4-5-deployment_rates_chat_web_search_two_requests_stream"}], + "messages": [ + { + "role": "user", + "content": "azure-ai-claude-sonnet-4-5-deployment_rates_chat_web_search_two_requests_stream", + } + ], "stream": True, "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 3}], "allowed_openai_params": ["web_search_options"], diff --git a/tests/integration/cost_calculation/cases/openai.py b/tests/integration/cost_calculation/cases/openai.py index ecf0af66079..eb7897ba259 100644 --- a/tests/integration/cost_calculation/cases/openai.py +++ b/tests/integration/cost_calculation/cases/openai.py @@ -5,6 +5,7 @@ PARITY pairs the streamed case with its plain twin where both bill the same row. from typing import Final, cast from integration.cost_calculation.cost_tracking_case import ( + with_usage, COST_MAP, BinaryResponse, CostTrackingTestCase, @@ -10620,6 +10621,226 @@ GPT_6_1_SOL_CHAT_BRIDGED_JSON: Final = GPT_6_1_SOL_RESPONSES_DEPLOYMENT_JSON.mod } ) +GPT_6_SOL_TIERED_INPUT_ABOVE_272K: Final = with_usage( + GPT_6_SOL_INPUT_TEXT, + "gpt-6-sol-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=1.329064, input_cost=1.32, output_cost=0.009064, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_SOL_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + GPT_6_SOL_INPUT_TEXT, + "gpt-6-sol-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=0.537064, input_cost=0.528, output_cost=0.009064, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_SOL_TIERED_AT_272K_BOUNDARY: Final = with_usage( + GPT_6_SOL_INPUT_TEXT, + "gpt-6-sol-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.602932, input_cost=0.5984, output_cost=0.004532, prompt_tokens=272000, completion_tokens=412), +) + +GPT_6_SOL_TIERED_FLEX_ABOVE_272K: Final = with_usage( + GPT_6_SOL_SERVICE_TIER_FLEX, + "gpt-6-sol-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.664532, input_cost=0.66, output_cost=0.004532, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_SOL_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + GPT_6_SOL_SERVICE_TIER_PRIORITY, + "gpt-6-sol-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=2.658128, input_cost=2.64, output_cost=0.018128, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_LUNA_TIERED_INPUT_ABOVE_272K: Final = with_usage( + GPT_6_LUNA_INPUT_TEXT, + "gpt-6-luna-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.060412, input_cost=0.06, output_cost=0.000412, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_LUNA_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + GPT_6_LUNA_INPUT_TEXT, + "gpt-6-luna-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=0.024412, input_cost=0.024, output_cost=0.000412, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_LUNA_TIERED_AT_272K_BOUNDARY: Final = with_usage( + GPT_6_LUNA_INPUT_TEXT, + "gpt-6-luna-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.027406, input_cost=0.0272, output_cost=0.000206, prompt_tokens=272000, completion_tokens=412), +) + +GPT_6_LUNA_TIERED_FLEX_ABOVE_272K: Final = with_usage( + GPT_6_LUNA_SERVICE_TIER_FLEX, + "gpt-6-luna-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.030206, input_cost=0.03, output_cost=0.000206, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_LUNA_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + GPT_6_LUNA_SERVICE_TIER_PRIORITY, + "gpt-6-luna-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.120824, input_cost=0.12, output_cost=0.000824, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_ASTRA_TIERED_INPUT_ABOVE_272K: Final = with_usage( + GPT_6_ASTRA_INPUT_TEXT, + "gpt-6-astra-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=7.24944, input_cost=7.2, output_cost=0.04944, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_ASTRA_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + GPT_6_ASTRA_INPUT_TEXT, + "gpt-6-astra-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=2.92944, input_cost=2.88, output_cost=0.04944, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_ASTRA_TIERED_AT_272K_BOUNDARY: Final = with_usage( + GPT_6_ASTRA_INPUT_TEXT, + "gpt-6-astra-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=3.28872, input_cost=3.264, output_cost=0.02472, prompt_tokens=272000, completion_tokens=412), +) + +GPT_6_ASTRA_TIERED_FLEX_ABOVE_272K: Final = with_usage( + GPT_6_ASTRA_SERVICE_TIER_FLEX, + "gpt-6-astra-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=3.62472, input_cost=3.6, output_cost=0.02472, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_ASTRA_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + GPT_6_ASTRA_SERVICE_TIER_PRIORITY, + "gpt-6-astra-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=14.49888, input_cost=14.4, output_cost=0.09888, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_1_SOL_TIERED_INPUT_ABOVE_272K: Final = with_usage( + GPT_6_1_SOL_INPUT_TEXT, + "gpt-6.1-sol-tiered_input_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=1.20824, input_cost=1.2, output_cost=0.00824, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_1_SOL_TIERED_CACHE_READ_ABOVE_272K: Final = with_usage( + GPT_6_1_SOL_INPUT_TEXT, + "gpt-6.1-sol-tiered_cache_read_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 200000}, + }, + ExactExpected(spend=0.44824, input_cost=0.44, output_cost=0.00824, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_1_SOL_TIERED_AT_272K_BOUNDARY: Final = with_usage( + GPT_6_1_SOL_INPUT_TEXT, + "gpt-6.1-sol-tiered_at_272k_boundary", + { + "prompt_tokens": 272000, + "total_tokens": 272412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.54812, input_cost=0.544, output_cost=0.00412, prompt_tokens=272000, completion_tokens=412), +) + +GPT_6_1_SOL_TIERED_FLEX_ABOVE_272K: Final = with_usage( + GPT_6_1_SOL_SERVICE_TIER_FLEX, + "gpt-6.1-sol-tiered_flex_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=0.60412, input_cost=0.6, output_cost=0.00412, prompt_tokens=300000, completion_tokens=412), +) + +GPT_6_1_SOL_TIERED_PRIORITY_ABOVE_272K: Final = with_usage( + GPT_6_1_SOL_SERVICE_TIER_PRIORITY, + "gpt-6.1-sol-tiered_priority_above_272k", + { + "prompt_tokens": 300000, + "total_tokens": 300412, + "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0}, + }, + ExactExpected(spend=2.41648, input_cost=2.4, output_cost=0.01648, prompt_tokens=300000, completion_tokens=412), +) + CASES: Final[tuple[CostTrackingTestCase, ...]] = ( GPT_6_SOL_RESPONSES_DEPLOYMENT_JSON, GPT_6_SOL_CHAT_BRIDGED_JSON, @@ -10629,6 +10850,26 @@ CASES: Final[tuple[CostTrackingTestCase, ...]] = ( GPT_6_ASTRA_CHAT_BRIDGED_JSON, GPT_6_1_SOL_RESPONSES_DEPLOYMENT_JSON, GPT_6_1_SOL_CHAT_BRIDGED_JSON, + GPT_6_SOL_TIERED_INPUT_ABOVE_272K, + GPT_6_SOL_TIERED_CACHE_READ_ABOVE_272K, + GPT_6_SOL_TIERED_AT_272K_BOUNDARY, + GPT_6_SOL_TIERED_FLEX_ABOVE_272K, + GPT_6_SOL_TIERED_PRIORITY_ABOVE_272K, + GPT_6_LUNA_TIERED_INPUT_ABOVE_272K, + GPT_6_LUNA_TIERED_CACHE_READ_ABOVE_272K, + GPT_6_LUNA_TIERED_AT_272K_BOUNDARY, + GPT_6_LUNA_TIERED_FLEX_ABOVE_272K, + GPT_6_LUNA_TIERED_PRIORITY_ABOVE_272K, + GPT_6_ASTRA_TIERED_INPUT_ABOVE_272K, + GPT_6_ASTRA_TIERED_CACHE_READ_ABOVE_272K, + GPT_6_ASTRA_TIERED_AT_272K_BOUNDARY, + GPT_6_ASTRA_TIERED_FLEX_ABOVE_272K, + GPT_6_ASTRA_TIERED_PRIORITY_ABOVE_272K, + GPT_6_1_SOL_TIERED_INPUT_ABOVE_272K, + GPT_6_1_SOL_TIERED_CACHE_READ_ABOVE_272K, + GPT_6_1_SOL_TIERED_AT_272K_BOUNDARY, + GPT_6_1_SOL_TIERED_FLEX_ABOVE_272K, + GPT_6_1_SOL_TIERED_PRIORITY_ABOVE_272K, GPT_5_3_CODEX_INPUT_TEXT, GPT_5_3_CODEX_CACHE_READ, GPT_5_3_CODEX_REASONING, diff --git a/tests/integration/cost_calculation/cost_map.py b/tests/integration/cost_calculation/cost_map.py index ac51f96a0ec..85388dc05b5 100644 --- a/tests/integration/cost_calculation/cost_map.py +++ b/tests/integration/cost_calculation/cost_map.py @@ -54,8 +54,12 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "mode": "chat", "output_cost_per_token": 5e-07, "output_cost_per_token_decisions": 0.0, - "output_cost_per_token_above_272k_tokens": 7.5e-07, + "output_cost_per_token_above_272k_tokens": 1e-06, "output_cost_per_token_above_272k_tokens_decisions": 0.0, + "input_cost_per_token_above_272k_tokens_flex": 1e-07, + "input_cost_per_token_above_272k_tokens_priority": 4e-07, + "output_cost_per_token_above_272k_tokens_flex": 5e-07, + "output_cost_per_token_above_272k_tokens_priority": 2e-06, }, "perplexity/sonar-next": { "litellm_provider": "perplexity", @@ -836,6 +840,13 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "max_input_tokens": 922000, "max_output_tokens": 128000, "supports_function_calling": True, + "input_cost_per_token_above_272k_tokens": 4e-06, + "input_cost_per_token_above_272k_tokens_flex": 2e-06, + "input_cost_per_token_above_272k_tokens_priority": 8e-06, + "output_cost_per_token_above_272k_tokens": 2e-05, + "output_cost_per_token_above_272k_tokens_flex": 1e-05, + "output_cost_per_token_above_272k_tokens_priority": 4e-05, + "cache_read_input_token_cost_above_272k_tokens": 2e-07, }, "claude-sonnet-5-5": { "cache_creation_input_token_cost": 4.125e-06, @@ -862,6 +873,11 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "mode": "chat", "output_cost_per_token": 5.5e-06, "supports_function_calling": True, + "input_cost_per_token_above_100k_tokens": 2.2e-06, + "output_cost_per_token_above_100k_tokens": 1.1e-05, + "cache_read_input_token_cost_above_100k_tokens": 2.2e-07, + "cache_creation_input_token_cost_above_100k_tokens": 2.75e-06, + "cache_creation_input_token_cost_above_1hr_above_100k_tokens": 4.4e-06, }, "claude-fable-5": { "cache_creation_input_token_cost": 1.875e-05, @@ -940,6 +956,11 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "mode": "chat", "output_cost_per_token": 6e-06, "supports_function_calling": True, + "input_cost_per_token_above_100k_tokens": 2.4e-06, + "output_cost_per_token_above_100k_tokens": 1.2e-05, + "cache_read_input_token_cost_above_100k_tokens": 2.4e-07, + "cache_creation_input_token_cost_above_100k_tokens": 3e-06, + "cache_creation_input_token_cost_above_1hr_above_100k_tokens": 4.8e-06, }, "us.anthropic.claude-fable-5": { "cache_creation_input_token_cost": 2.125e-05, @@ -981,6 +1002,13 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "output_cost_per_token_flex": 5.5e-06, "output_cost_per_token_priority": 2.2e-05, "supports_function_calling": True, + "input_cost_per_token_above_272k_tokens": 4.4e-06, + "input_cost_per_token_above_272k_tokens_flex": 2.2e-06, + "input_cost_per_token_above_272k_tokens_priority": 8.8e-06, + "output_cost_per_token_above_272k_tokens": 2.2e-05, + "output_cost_per_token_above_272k_tokens_flex": 1.1e-05, + "output_cost_per_token_above_272k_tokens_priority": 4.4e-05, + "cache_read_input_token_cost_above_272k_tokens": 4.4e-07, }, "gpt-6-astra": { "cache_read_input_token_cost": 1.2e-06, @@ -996,6 +1024,13 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "output_cost_per_token_flex": 3e-05, "output_cost_per_token_priority": 0.00012, "supports_function_calling": True, + "input_cost_per_token_above_272k_tokens": 2.4e-05, + "input_cost_per_token_above_272k_tokens_flex": 1.2e-05, + "input_cost_per_token_above_272k_tokens_priority": 4.8e-05, + "output_cost_per_token_above_272k_tokens": 0.00012, + "output_cost_per_token_above_272k_tokens_flex": 6e-05, + "output_cost_per_token_above_272k_tokens_priority": 0.00024, + "cache_read_input_token_cost_above_272k_tokens": 2.4e-06, }, "azure/gpt-6-sol": { "cache_read_input_token_cost": 2.3e-07, @@ -1012,6 +1047,13 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "output_cost_per_token_priority": 2.3e-05, "supports_function_calling": True, "supports_none_reasoning_effort": True, + "input_cost_per_token_above_272k_tokens": 4.6e-06, + "input_cost_per_token_above_272k_tokens_flex": 2.3e-06, + "input_cost_per_token_above_272k_tokens_priority": 9.2e-06, + "output_cost_per_token_above_272k_tokens": 2.3e-05, + "output_cost_per_token_above_272k_tokens_flex": 1.15e-05, + "output_cost_per_token_above_272k_tokens_priority": 4.6e-05, + "cache_read_input_token_cost_above_272k_tokens": 4.6e-07, }, "azure/gpt-6-luna": { "cache_read_input_token_cost": 1.3e-08, @@ -1028,6 +1070,13 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "output_cost_per_token_priority": 1.3e-06, "supports_function_calling": True, "supports_none_reasoning_effort": True, + "input_cost_per_token_above_272k_tokens": 2.6e-07, + "input_cost_per_token_above_272k_tokens_flex": 1.3e-07, + "input_cost_per_token_above_272k_tokens_priority": 5.2e-07, + "output_cost_per_token_above_272k_tokens": 1.3e-06, + "output_cost_per_token_above_272k_tokens_flex": 6.5e-07, + "output_cost_per_token_above_272k_tokens_priority": 2.6e-06, + "cache_read_input_token_cost_above_272k_tokens": 2.6e-08, }, "azure/gpt-6-astra": { "cache_read_input_token_cost": 1.25e-06, @@ -1044,6 +1093,13 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "output_cost_per_token_priority": 0.000125, "supports_function_calling": True, "supports_none_reasoning_effort": True, + "input_cost_per_token_above_272k_tokens": 2.5e-05, + "input_cost_per_token_above_272k_tokens_flex": 1.25e-05, + "input_cost_per_token_above_272k_tokens_priority": 5e-05, + "output_cost_per_token_above_272k_tokens": 0.000125, + "output_cost_per_token_above_272k_tokens_flex": 6.25e-05, + "output_cost_per_token_above_272k_tokens_priority": 0.00025, + "cache_read_input_token_cost_above_272k_tokens": 2.5e-06, }, "azure/gpt-6.1-sol": { "cache_read_input_token_cost": 2.4e-07, @@ -1060,5 +1116,12 @@ COST_MAP_ENTRIES: Final[dict[str, dict[str, JsonValue]]] = { "output_cost_per_token_priority": 2.4e-05, "supports_function_calling": True, "supports_none_reasoning_effort": True, + "input_cost_per_token_above_272k_tokens": 4.8e-06, + "input_cost_per_token_above_272k_tokens_flex": 2.4e-06, + "input_cost_per_token_above_272k_tokens_priority": 9.6e-06, + "output_cost_per_token_above_272k_tokens": 2.4e-05, + "output_cost_per_token_above_272k_tokens_flex": 1.2e-05, + "output_cost_per_token_above_272k_tokens_priority": 4.8e-05, + "cache_read_input_token_cost_above_272k_tokens": 4.8e-07, }, } diff --git a/tests/integration/cost_calculation/cost_tracking_case.py b/tests/integration/cost_calculation/cost_tracking_case.py index 714c2bf3730..52319a79beb 100644 --- a/tests/integration/cost_calculation/cost_tracking_case.py +++ b/tests/integration/cost_calculation/cost_tracking_case.py @@ -44,6 +44,15 @@ class CostMapEntry(BaseModel): max_output_tokens: int | None = None supports_function_calling: bool | None = None supports_none_reasoning_effort: bool | None = None + cache_creation_input_token_cost_above_100k_tokens: float | None = None + cache_creation_input_token_cost_above_1hr_above_100k_tokens: float | None = None + cache_read_input_token_cost_above_100k_tokens: float | None = None + input_cost_per_token_above_100k_tokens: float | None = None + input_cost_per_token_above_272k_tokens_flex: float | None = None + input_cost_per_token_above_272k_tokens_priority: float | None = None + output_cost_per_token_above_100k_tokens: float | None = None + output_cost_per_token_above_272k_tokens_flex: float | None = None + output_cost_per_token_above_272k_tokens_priority: float | None = None input_cost_per_token: float | None = None input_cost_per_token_decisions: float | None = None input_cost_per_query: float | None = None @@ -587,3 +596,22 @@ _LITELLM_PARAMS: Final[Mapping[str, Mapping[str, str]]] = MappingProxyType( COST_MAP: Final[Mapping[str, CostMapEntry]] = MappingProxyType( {model: CostMapEntry.model_validate(entry) for model, entry in COST_MAP_ENTRIES.items()} ) + + +def with_usage( + base: CostTrackingTestCase, name: str, usage: Mapping[str, JsonValue], expected: Expected +) -> CostTrackingTestCase: + """Same request and stored body as ``base``, with the provider usage block and the expected row swapped.""" + if not isinstance(base.response, JsonResponse): + raise TypeError(f"{base.name}: with_usage needs a JsonResponse base") + body: Final = base.response.body + current: Final = body["usage"] + if not isinstance(current, dict): + raise TypeError(f"{base.name}: stored body has no usage object") + return base.model_copy( + update={ + "name": name, + "response": base.response.model_copy(update={"body": {**body, "usage": {**current, **usage}}}), + "expected": expected, + } + )