scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-08-28 05:25:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 36

Merge pull request #37197 from BerriAI/litellm_gemini_36_flash_introductory_pricing

Browse source 
fix(gemini): price gemini 3.6 flash at Google's introductory rates on every service tier
This commit is contained in:
Mateo Wang 2026-08-17 15:32:08 -07:00 • committed by GitHub
commit ce20c282e9
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 122 additions and 77 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

72
litellm/model_prices_and_context_window_backup.json
View file

@ -19424,20 +19424,20 @@
"web_search_billing_unit": "per_query"
},
"vertex_ai/gemini-3.6-flash": {
"cache_read_input_token_cost": 1.5e-07,
"cache_read_input_token_cost_flex": 7.5e-08,
"input_cost_per_token": 1.5e-06,
"input_cost_per_token_batches": 7.5e-07,
"input_cost_per_token_flex": 7.5e-07,
"cache_read_input_token_cost": 7.5e-08,
"cache_read_input_token_cost_flex": 3.75e-08,
"input_cost_per_token": 7.5e-07,
"input_cost_per_token_batches": 3.75e-07,
"input_cost_per_token_flex": 3.75e-07,
"litellm_provider": "vertex_ai",
"max_input_tokens": 1048576,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 7.5e-06,
"output_cost_per_token": 7.5e-06,
"output_cost_per_token_batches": 3.75e-06,
"output_cost_per_token_flex": 3.75e-06,
"output_cost_per_reasoning_token": 3.75e-06,
"output_cost_per_token": 3.75e-06,
"output_cost_per_token_batches": 1.875e-06,
"output_cost_per_token_flex": 1.875e-06,
"source": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
"supported_endpoints": [
"/v1/chat/completions",
@ -19467,9 +19467,9 @@
"supports_vision": true,
"supports_web_search": true,
"supports_native_streaming": true,
"input_cost_per_token_priority": 2.7e-06,
"output_cost_per_token_priority": 1.35e-05,
"cache_read_input_token_cost_priority": 2.7e-07,
"input_cost_per_token_priority": 1.35e-06,
"output_cost_per_token_priority": 6.75e-06,
"cache_read_input_token_cost_priority": 1.35e-07,
"search_context_cost_per_query": {
"search_context_size_low": 0.014,
"search_context_size_medium": 0.014,
@ -21150,20 +21150,20 @@
"web_search_billing_unit": "per_query"
},
"gemini/gemini-3.6-flash": {
"cache_read_input_token_cost": 1.5e-07,
"cache_read_input_token_cost_flex": 7.5e-08,
"input_cost_per_token": 1.5e-06,
"input_cost_per_token_batches": 7.5e-07,
"input_cost_per_token_flex": 7.5e-07,
"cache_read_input_token_cost": 7.5e-08,
"cache_read_input_token_cost_flex": 3.75e-08,
"input_cost_per_token": 7.5e-07,
"input_cost_per_token_batches": 3.75e-07,
"input_cost_per_token_flex": 3.75e-07,
"litellm_provider": "gemini",
"max_input_tokens": 1048576,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 7.5e-06,
"output_cost_per_token": 7.5e-06,
"output_cost_per_token_batches": 3.75e-06,
"output_cost_per_token_flex": 3.75e-06,
"output_cost_per_reasoning_token": 3.75e-06,
"output_cost_per_token": 3.75e-06,
"output_cost_per_token_batches": 1.875e-06,
"output_cost_per_token_flex": 1.875e-06,
"rpm": 2000,
"source": "https://ai.google.dev/pricing/gemini-3",
"supported_endpoints": [
@ -21196,9 +21196,9 @@
"supports_web_search": true,
"supports_native_streaming": true,
"tpm": 800000,
"input_cost_per_token_priority": 2.7e-06,
"output_cost_per_token_priority": 1.35e-05,
"cache_read_input_token_cost_priority": 2.7e-07,
"input_cost_per_token_priority": 1.35e-06,
"output_cost_per_token_priority": 6.75e-06,
"cache_read_input_token_cost_priority": 1.35e-07,
"search_context_cost_per_query": {
"search_context_size_low": 0.014,
"search_context_size_medium": 0.014,
@ -21544,20 +21544,20 @@
"web_search_billing_unit": "per_query"
},
"gemini-3.6-flash": {
"cache_read_input_token_cost": 1.5e-07,
"cache_read_input_token_cost_flex": 7.5e-08,
"input_cost_per_token": 1.5e-06,
"input_cost_per_token_batches": 7.5e-07,
"input_cost_per_token_flex": 7.5e-07,
"cache_read_input_token_cost": 7.5e-08,
"cache_read_input_token_cost_flex": 3.75e-08,
"input_cost_per_token": 7.5e-07,
"input_cost_per_token_batches": 3.75e-07,
"input_cost_per_token_flex": 3.75e-07,
"litellm_provider": "vertex_ai-language-models",
"max_input_tokens": 1048576,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 7.5e-06,
"output_cost_per_token": 7.5e-06,
"output_cost_per_token_batches": 3.75e-06,
"output_cost_per_token_flex": 3.75e-06,
"output_cost_per_reasoning_token": 3.75e-06,
"output_cost_per_token": 3.75e-06,
"output_cost_per_token_batches": 1.875e-06,
"output_cost_per_token_flex": 1.875e-06,
"source": "https://ai.google.dev/pricing/gemini-3",
"supported_endpoints": [
"/v1/chat/completions",
@ -21588,9 +21588,9 @@
"supports_vision": true,
"supports_web_search": true,
"supports_native_streaming": true,
"input_cost_per_token_priority": 2.7e-06,
"output_cost_per_token_priority": 1.35e-05,
"cache_read_input_token_cost_priority": 2.7e-07,
"input_cost_per_token_priority": 1.35e-06,
"output_cost_per_token_priority": 6.75e-06,
"cache_read_input_token_cost_priority": 1.35e-07,
"search_context_cost_per_query": {
"search_context_size_low": 0.014,
"search_context_size_medium": 0.014,

72
model_prices_and_context_window.json
View file

@ -19424,20 +19424,20 @@
"web_search_billing_unit": "per_query"
},
"vertex_ai/gemini-3.6-flash": {
"cache_read_input_token_cost": 1.5e-07,
"cache_read_input_token_cost_flex": 7.5e-08,
"input_cost_per_token": 1.5e-06,
"input_cost_per_token_batches": 7.5e-07,
"input_cost_per_token_flex": 7.5e-07,
"cache_read_input_token_cost": 7.5e-08,
"cache_read_input_token_cost_flex": 3.75e-08,
"input_cost_per_token": 7.5e-07,
"input_cost_per_token_batches": 3.75e-07,
"input_cost_per_token_flex": 3.75e-07,
"litellm_provider": "vertex_ai",
"max_input_tokens": 1048576,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 7.5e-06,
"output_cost_per_token": 7.5e-06,
"output_cost_per_token_batches": 3.75e-06,
"output_cost_per_token_flex": 3.75e-06,
"output_cost_per_reasoning_token": 3.75e-06,
"output_cost_per_token": 3.75e-06,
"output_cost_per_token_batches": 1.875e-06,
"output_cost_per_token_flex": 1.875e-06,
"source": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
"supported_endpoints": [
"/v1/chat/completions",
@ -19467,9 +19467,9 @@
"supports_vision": true,
"supports_web_search": true,
"supports_native_streaming": true,
"input_cost_per_token_priority": 2.7e-06,
"output_cost_per_token_priority": 1.35e-05,
"cache_read_input_token_cost_priority": 2.7e-07,
"input_cost_per_token_priority": 1.35e-06,
"output_cost_per_token_priority": 6.75e-06,
"cache_read_input_token_cost_priority": 1.35e-07,
"search_context_cost_per_query": {
"search_context_size_low": 0.014,
"search_context_size_medium": 0.014,
@ -21150,20 +21150,20 @@
"web_search_billing_unit": "per_query"
},
"gemini/gemini-3.6-flash": {
"cache_read_input_token_cost": 1.5e-07,
"cache_read_input_token_cost_flex": 7.5e-08,
"input_cost_per_token": 1.5e-06,
"input_cost_per_token_batches": 7.5e-07,
"input_cost_per_token_flex": 7.5e-07,
"cache_read_input_token_cost": 7.5e-08,
"cache_read_input_token_cost_flex": 3.75e-08,
"input_cost_per_token": 7.5e-07,
"input_cost_per_token_batches": 3.75e-07,
"input_cost_per_token_flex": 3.75e-07,
"litellm_provider": "gemini",
"max_input_tokens": 1048576,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 7.5e-06,
"output_cost_per_token": 7.5e-06,
"output_cost_per_token_batches": 3.75e-06,
"output_cost_per_token_flex": 3.75e-06,
"output_cost_per_reasoning_token": 3.75e-06,
"output_cost_per_token": 3.75e-06,
"output_cost_per_token_batches": 1.875e-06,
"output_cost_per_token_flex": 1.875e-06,
"rpm": 2000,
"source": "https://ai.google.dev/pricing/gemini-3",
"supported_endpoints": [
@ -21196,9 +21196,9 @@
"supports_web_search": true,
"supports_native_streaming": true,
"tpm": 800000,
"input_cost_per_token_priority": 2.7e-06,
"output_cost_per_token_priority": 1.35e-05,
"cache_read_input_token_cost_priority": 2.7e-07,
"input_cost_per_token_priority": 1.35e-06,
"output_cost_per_token_priority": 6.75e-06,
"cache_read_input_token_cost_priority": 1.35e-07,
"search_context_cost_per_query": {
"search_context_size_low": 0.014,
"search_context_size_medium": 0.014,
@ -21544,20 +21544,20 @@
"web_search_billing_unit": "per_query"
},
"gemini-3.6-flash": {
"cache_read_input_token_cost": 1.5e-07,
"cache_read_input_token_cost_flex": 7.5e-08,
"input_cost_per_token": 1.5e-06,
"input_cost_per_token_batches": 7.5e-07,
"input_cost_per_token_flex": 7.5e-07,
"cache_read_input_token_cost": 7.5e-08,
"cache_read_input_token_cost_flex": 3.75e-08,
"input_cost_per_token": 7.5e-07,
"input_cost_per_token_batches": 3.75e-07,
"input_cost_per_token_flex": 3.75e-07,
"litellm_provider": "vertex_ai-language-models",
"max_input_tokens": 1048576,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 7.5e-06,
"output_cost_per_token": 7.5e-06,
"output_cost_per_token_batches": 3.75e-06,
"output_cost_per_token_flex": 3.75e-06,
"output_cost_per_reasoning_token": 3.75e-06,
"output_cost_per_token": 3.75e-06,
"output_cost_per_token_batches": 1.875e-06,
"output_cost_per_token_flex": 1.875e-06,
"source": "https://ai.google.dev/pricing/gemini-3",
"supported_endpoints": [
"/v1/chat/completions",
@ -21588,9 +21588,9 @@
"supports_vision": true,
"supports_web_search": true,
"supports_native_streaming": true,
"input_cost_per_token_priority": 2.7e-06,
"output_cost_per_token_priority": 1.35e-05,
"cache_read_input_token_cost_priority": 2.7e-07,
"input_cost_per_token_priority": 1.35e-06,
"output_cost_per_token_priority": 6.75e-06,
"cache_read_input_token_cost_priority": 1.35e-07,
"search_context_cost_per_query": {
"search_context_size_low": 0.014,
"search_context_size_medium": 0.014,

55
tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
View file

@ -2922,9 +2922,9 @@ def test_image_response_input_image_tokens_priced_at_image_rate(details_as_dict)
assert cost is not None
assert round(cost, 12) == round(expected, 12)
GEMINI_DAY0_LAUNCH_PRICING = [
("gemini-3.6-flash", 1.5e-06, 7.5e-06, 1.5e-07),
("gemini/gemini-3.6-flash", 1.5e-06, 7.5e-06, 1.5e-07),
("vertex_ai/gemini-3.6-flash", 1.5e-06, 7.5e-06, 1.5e-07),
("gemini-3.6-flash", 7.5e-07, 3.75e-06, 7.5e-08),
("gemini/gemini-3.6-flash", 7.5e-07, 3.75e-06, 7.5e-08),
("vertex_ai/gemini-3.6-flash", 7.5e-07, 3.75e-06, 7.5e-08),
("gemini-3.5-flash-lite", 3e-07, 2.5e-06, 3e-08),
("gemini/gemini-3.5-flash-lite", 3e-07, 2.5e-06, 3e-08),
("vertex_ai/gemini-3.5-flash-lite", 3e-07, 2.5e-06, 3e-08),
@ -2966,8 +2966,53 @@ def test_generic_cost_per_token_gemini_36_flash():
usage=usage,
custom_llm_provider="gemini",
)
assert prompt_cost == pytest.approx(0.0015)
assert completion_cost == pytest.approx(0.00375)
assert prompt_cost == pytest.approx(0.00075)
assert completion_cost == pytest.approx(0.001875)
GEMINI_36_FLASH_SERVICE_TIER_PRICING = [
(None, 7.5e-07, 3.75e-06, 7.5e-08),
("flex", 3.75e-07, 1.875e-06, 3.75e-08),
("priority", 1.35e-06, 6.75e-06, 1.35e-07),
]
@pytest.mark.parametrize(
"service_tier,input_rate,output_rate,cache_read_rate", GEMINI_36_FLASH_SERVICE_TIER_PRICING
)
@pytest.mark.parametrize(
"model", ["gemini-3.6-flash", "gemini/gemini-3.6-flash", "vertex_ai/gemini-3.6-flash"]
)
def test_gemini_36_flash_service_tier_introductory_pricing(
model, service_tier, input_rate, output_rate, cache_read_rate, _local_model_cost_map
):
"""Regression: every 3.6 Flash tier is on Google's introductory rates through 2026-12-31,
so flex and priority requests must not be billed at the post-introductory rates."""
usage = Usage(
prompt_tokens=1_000,
completion_tokens=500,
total_tokens=1_500,
prompt_tokens_details=PromptTokensDetailsWrapper(cached_tokens=200, text_tokens=800),
)
prompt_cost, completion_cost = generic_cost_per_token(
model=model.split("/")[-1],
usage=usage,
custom_llm_provider=model.split("/")[0] if "/" in model else "gemini",
service_tier=service_tier,
)
assert prompt_cost == pytest.approx(800 * input_rate + 200 * cache_read_rate, rel=1e-9)
assert completion_cost == pytest.approx(500 * output_rate, rel=1e-9)
@pytest.mark.parametrize(
"model", ["gemini-3.6-flash", "gemini/gemini-3.6-flash", "vertex_ai/gemini-3.6-flash"]
)
def test_gemini_36_flash_batch_introductory_pricing(model, _local_model_cost_map):
model_cost_map = litellm.model_cost[model]
assert model_cost_map["input_cost_per_token_batches"] == 3.75e-07
assert model_cost_map["output_cost_per_token_batches"] == 1.875e-06
def test_generic_cost_per_token_gemini_35_flash_lite():