scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-19 00:01:29 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 88

Keep generated pricing parameters immutable

Browse source
This commit is contained in:
Yuneng Jiang 2026-09-14 21:17:12 -07:00
parent 2f65ef0ea3
commit a478a46d0a
No known key found for this signature in database
1 changed files with 60 additions and 19 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

79
tests/integration/pricing/test_price_precedence.py
View file

@ -16,20 +16,36 @@ def test_generated_zero_null_and_omitted_prices_follow_independent_arithmetic(ga
@example(rates=(0, 0))
@example(rates=(1, 2))
@example(rates=("null", "null"))
@given(rates=st.one_of(st.sampled_from((("omitted", "omitted"), ("null", "null"))), st.tuples(st.integers(0, 25), st.integers(0, 25))))
@given(
rates=st.one_of(
st.sampled_from((("omitted", "omitted"), ("null", "null"))),
st.tuples(st.integers(0, 25), st.integers(0, 25)),
)
)
def check(rates: tuple[str | int, str | int]) -> None:
if rates[0] in ("omitted", "null"):
parameters: Final = {} if rates[0] == "omitted" else {"input_cost_per_token": None, "output_cost_per_token": None}
input_rate, output_rate = 0.00000015, 0.0000006
else:
assert isinstance(rates[0], int) and isinstance(rates[1], int)
input_rate, output_rate = rates[0] / 1_000_000, rates[1] / 1_000_000
parameters = {"input_cost_per_token": input_rate, "output_cost_per_token": output_rate}
defaults: Final = rates[0] in ("omitted", "null")
assert defaults or (isinstance(rates[0], int) and isinstance(rates[1], int))
input_rate, output_rate = (
(0.00000015, 0.0000006) if defaults else (float(rates[0]) / 1_000_000, float(rates[1]) / 1_000_000)
)
parameters: Final = (
{}
if rates[0] == "omitted"
else {
"input_cost_per_token": None if rates[0] == "null" else input_rate,
"output_cost_per_token": None if rates[0] == "null" else output_rate,
}
)
with gateway.scenario() as scenario:
model: Final = scenario.model(**parameters)
response: Final = gateway.request("POST", "/v1/chat/completions", {
"model": model, "messages": [{"role": "user", "content": f"independent price {uuid.uuid4().hex}"}],
})
response: Final = gateway.request(
"POST",
"/v1/chat/completions",
{
"model": model,
"messages": [{"role": "user", "content": f"independent price {uuid.uuid4().hex}"}],
},
)
assert response.status_code == 200, response.text
assert response.json()["usage"] == {"prompt_tokens": 20, "completion_tokens": 20, "total_tokens": 40}
expected: Final = 20 * input_rate + 20 * output_rate
@ -37,10 +53,14 @@ def test_generated_zero_null_and_omitted_prices_follow_independent_arithmetic(ga
assert float(response.headers["x-litellm-response-cost"]) == pytest.approx(expected, rel=1e-6)
else:
assert response.headers.get("x-litellm-response-cost") in (None, "0", "0.0")
rows: Final = eventually(lambda: read_rows(
'SELECT spend, metadata, prompt_tokens, completion_tokens FROM "LiteLLM_SpendLogs" WHERE request_id=%s',
(response.json()["id"],),
), lambda values: len(values) == 1, seconds=70)
rows: Final = eventually(
lambda: read_rows(
'SELECT spend, metadata, prompt_tokens, completion_tokens FROM "LiteLLM_SpendLogs" WHERE request_id=%s',
(response.json()["id"],),
),
lambda values: len(values) == 1,
seconds=70,
)
assert rows[0]["prompt_tokens"] == 20 and rows[0]["completion_tokens"] == 20
assert float(rows[0]["spend"]) == pytest.approx(expected, rel=1e-6)
metadata: Final = rows[0]["metadata"]
@ -56,18 +76,35 @@ def test_generated_zero_null_and_omitted_prices_follow_independent_arithmetic(ga
def test_same_upstream_aliases_keep_distinct_prices_after_reload(gateway: Gateway) -> None:
for order in (("free", "paid"), ("paid", "free")):
with gateway.scenario() as scenario:
rates: Final = {"free": {"input_cost_per_token": 0, "output_cost_per_token": 0}, "paid": {"input_cost_per_token": 0.001, "output_cost_per_token": 0.003}}
rates: Final = {
"free": {"input_cost_per_token": 0, "output_cost_per_token": 0},
"paid": {"input_cost_per_token": 0.001, "output_cost_per_token": 0.003},
}
aliases: Final = {kind: scenario.model(**rates[kind]) for kind in order}
for generation in range(2):
for kind in order if generation == 0 else reversed(order):
model: Final = aliases[kind]
cost: Final = 0.08 if kind == "paid" else 0.0
response: Final = gateway.request("POST", "/v1/chat/completions", {"model": model, "messages": [{"role": "user", "content": f"alias price {model} {generation}"}]})
response: Final = gateway.request(
"POST",
"/v1/chat/completions",
{
"model": model,
"messages": [{"role": "user", "content": f"alias price {model} {generation}"}],
},
)
assert response.status_code == 200, response.text
assert response.json()["usage"]["total_tokens"] == 40
if cost:
assert float(response.headers["x-litellm-response-cost"]) == pytest.approx(cost)
rows: Final = eventually(lambda response=response: read_rows('SELECT spend, metadata FROM "LiteLLM_SpendLogs" WHERE request_id=%s', (response.json()["id"],)), lambda values: len(values) == 1, seconds=70)
rows: Final = eventually(
lambda response=response: read_rows(
'SELECT spend, metadata FROM "LiteLLM_SpendLogs" WHERE request_id=%s',
(response.json()["id"],),
),
lambda values: len(values) == 1,
seconds=70,
)
assert float(rows[0]["spend"]) == pytest.approx(cost)
metadata: Final = rows[0]["metadata"]
parsed: Final = json.loads(metadata) if isinstance(metadata, str) else object_value(metadata)
@ -77,5 +114,9 @@ def test_same_upstream_aliases_keep_distinct_prices_after_reload(gateway: Gatewa
if generation == 0:
entries: Final = gateway.get("/model/info")["data"]
target: Final = next(entry for entry in entries if entry["model_name"] == aliases["paid"])
changed: Final = gateway.request("PATCH", f"/model/{target['model_info']['id']}/update", {"model_info": {"description": "price reload"}})
changed: Final = gateway.request(
"PATCH",
f"/model/{target['model_info']['id']}/update",
{"model_info": {"description": "price reload"}},
)
assert changed.status_code == 200, changed.text