mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-12 03:48:56 +00:00
ee295aea68
* refactor(rust): give Messages configs typed litellm params * feat(rust): port _normalize_system_role_messages for Messages hosts mid_conversation_system mirrors Python's module function for function: a leading run of system turns is hoisted into the top-level system, a later turn stays in place when the model map flags supports_mid_conversation_system and otherwise becomes a user turn where it was, never between a tool_use and its tool_result, and billing blocks are stripped from the hoisted system. The capability joins MessagesModelCapabilities and the bridge projection. Azure AI uses it in place of fold_system_role_messages, which hoisted every turn. * fix(rust): keep the Bedrock runtime endpoint behind a blank api_base * refactor(rust): import LitellmParams from its crate instead of a re-export * refactor(rust): template the Messages request decode errors Both decode failures go through ErrorDetail::invalid with the subject and the serde error as source instead of a preformatted sentence. * refactor(rust_bridge): read the module globals the param specs name The Messages host no longer keeps its own list of litellm globals; a spec whose spellings the call leaves out names the global to read, so the next family that has one needs no host change. * refactor(rust): give each route lookup Python's implicit None ProviderConfigManager's per-route methods list only the providers the route serves and fall through to None; the Rust lookups now do the same, so a new LlmProviders variant touches only the route that serves it.
76 lines
2.7 KiB
Python
76 lines
2.7 KiB
Python
from typing import Final
|
|
|
|
import pytest
|
|
|
|
import litellm
|
|
from litellm.rust_bridge.model_capabilities import anthropic_model_capabilities
|
|
|
|
|
|
pytestmark = pytest.mark.usefixtures("local_model_cost_map")
|
|
|
|
|
|
def _flag_model(monkeypatch: pytest.MonkeyPatch, name: str, **flags: bool) -> None:
|
|
monkeypatch.setitem(
|
|
litellm.model_cost,
|
|
name,
|
|
{
|
|
"litellm_provider": "anthropic",
|
|
"mode": "chat",
|
|
"input_cost_per_token": 0,
|
|
"output_cost_per_token": 0,
|
|
**flags,
|
|
},
|
|
)
|
|
|
|
|
|
def test_capabilities_come_from_the_model_map_under_the_callers_provider(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
_flag_model(
|
|
monkeypatch,
|
|
"claude-test-adaptive",
|
|
supports_reasoning=True,
|
|
supports_adaptive_thinking=True,
|
|
supports_output_config=True,
|
|
supports_xhigh_reasoning_effort=True,
|
|
supports_sampling_params=False,
|
|
supports_mid_conversation_system=True,
|
|
)
|
|
|
|
capabilities: Final = anthropic_model_capabilities("anthropic/claude-test-adaptive", None)
|
|
|
|
assert capabilities["supports_adaptive_thinking"]
|
|
assert capabilities["supports_output_config"]
|
|
assert capabilities["supports_mid_conversation_system"]
|
|
assert not capabilities["supports_legacy_thinking"]
|
|
assert not capabilities["supports_sampling_params"]
|
|
assert capabilities["effort_tiers"] == {
|
|
"minimal": False,
|
|
"low": False,
|
|
"medium": False,
|
|
"high": False,
|
|
"xhigh": True,
|
|
"max": False,
|
|
}
|
|
|
|
|
|
def test_unmapped_model_keeps_sampling_params_and_no_reasoning_features() -> None:
|
|
capabilities: Final = anthropic_model_capabilities("anthropic/not-a-real-model", None)
|
|
|
|
assert capabilities["supports_sampling_params"]
|
|
assert not capabilities["supports_reasoning"]
|
|
assert not capabilities["supports_adaptive_thinking"]
|
|
assert not capabilities["supports_mid_conversation_system"]
|
|
assert capabilities["effort_tiers"] == dict.fromkeys(("minimal", "low", "medium", "high", "xhigh", "max"), False)
|
|
|
|
|
|
def test_capability_source_observes_runtime_registration_changes(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
model: Final = "claude-test-runtime-registration"
|
|
_flag_model(monkeypatch, model, supports_reasoning=True, supports_output_config=True)
|
|
before: Final = anthropic_model_capabilities(model, "anthropic")
|
|
|
|
_flag_model(monkeypatch, model, supports_reasoning=False, supports_output_config=False)
|
|
after: Final = anthropic_model_capabilities(model, "anthropic")
|
|
|
|
assert before["supports_reasoning"] is True
|
|
assert before["supports_output_config"] is True
|
|
assert after["supports_reasoning"] is False
|
|
assert after["supports_output_config"] is False
|