Beto de Paola
a06f1d7617
muse-spark-1.2-contributor is heavily discounted BECAUSE Meta uses your prompts and completions to train future models. Selecting it for the price without realising the data trade-off is a footgun. Add hermes_cli/model_data_policy_guard.py (mirrors model_cost_guard): data_training_warning(model_id, provider, base_url) -> DataTrainingWarning|None, driven by a vendor-agnostic rule table. The status is not machine-readable on /v1/models or models.dev, so the v1 rule keys on the documented '-contributor' model id (fires regardless of provider, so it also covers custom/gateway routes). Message mirrors Meta's pricing-doc language and figures (https://dev.meta.ai/docs/pricing-rate-limits/). Wire it into the CLI model picker's confirm flow (auth.py) as a [y/N] disclosure, chained after the expensive-model cost guard. Fires only on the contributor tier; silent on muse-spark-1.1/1.2 and all other models.
39 lines
1.6 KiB
Python
39 lines
1.6 KiB
Python
"""Tests for the data-training-tier selection guard."""
|
|
|
|
from hermes_cli.model_data_policy_guard import (
|
|
DataTrainingWarning,
|
|
data_training_warning,
|
|
)
|
|
|
|
|
|
def test_fires_on_meta_contributor():
|
|
w = data_training_warning("muse-spark-1.2-contributor", provider="meta-ai")
|
|
assert isinstance(w, DataTrainingWarning)
|
|
assert w.model == "muse-spark-1.2-contributor"
|
|
assert "train" in w.message.lower()
|
|
assert "muse-spark-1.2" in w.message # points to the no-training alternative
|
|
# Aligns with Meta's own pricing doc language + figures.
|
|
assert "$0.10" in w.message and "$0.20" in w.message and "$0.002" in w.message
|
|
assert "prompts and completions" in w.message.lower()
|
|
assert "dev.meta.ai/docs/pricing-rate-limits" in w.message
|
|
|
|
|
|
def test_silent_on_non_contributor_muse():
|
|
assert data_training_warning("muse-spark-1.2", provider="meta-ai") is None
|
|
assert data_training_warning("muse-spark-1.1", provider="meta-ai") is None
|
|
|
|
|
|
def test_silent_on_unrelated_models():
|
|
for m in ("anthropic/claude-opus-4.8", "gpt-5.6-sol", "deepseek-v4-pro", ""):
|
|
assert data_training_warning(m, provider="anthropic") is None
|
|
|
|
|
|
def test_fires_regardless_of_provider_string():
|
|
# id-keyed: must fire even if selected via custom/gateway (no meta-ai provider)
|
|
assert data_training_warning("muse-spark-1.2-contributor", provider="custom") is not None
|
|
assert data_training_warning("muse-spark-1.2-contributor") is not None
|
|
|
|
|
|
def test_case_insensitive():
|
|
assert data_training_warning("MUSE-SPARK-1.2-CONTRIBUTOR", provider="meta-ai") is not None
|