mirror of
https://github.com/crewAIInc/crewAI.git
synced 2026-04-14 06:53:25 +00:00
feat(bedrock): add support for Anthropic Claude V4 models
- Add _strip_region_prefix() to handle us./eu./apac./global. prefixes - Update get_context_window_size() with V4 model entries (200K tokens) - Update supports_multimodal() to detect V4 models - Update _is_nova_model() to use region prefix stripping - Add V4 model context windows to LLM_CONTEXT_WINDOW_SIZES in llm.py - Add 11 comprehensive tests covering V4 model detection, context windows, multimodal support, region prefixes, function calling, and regression Closes #4700 Co-Authored-By: João <joao@crewai.com>
This commit is contained in:
Devin AI
@@ -248,6 +248,14 @@ LLM_CONTEXT_WINDOW_SIZES: Final[dict[str, int]] = {
|
||||
"us.amazon.nova-pro-v1:0": 300000,
|
||||
"us.amazon.nova-micro-v1:0": 128000,
|
||||
"us.amazon.nova-lite-v1:0": 300000,
|
||||
# Claude 4.x (us-prefixed)
|
||||
"us.anthropic.claude-sonnet-4-5-20250929-v1:0": 200000,
|
||||
"us.anthropic.claude-sonnet-4-20250514-v1:0": 200000,
|
||||
"us.anthropic.claude-opus-4-5-20251101-v1:0": 200000,
|
||||
"us.anthropic.claude-opus-4-1-20250805-v1:0": 200000,
|
||||
"us.anthropic.claude-opus-4-20250514-v1:0": 200000,
|
||||
"us.anthropic.claude-haiku-4-5-20251001-v1:0": 200000,
|
||||
# Claude 3.x (us-prefixed)
|
||||
"us.anthropic.claude-3-5-sonnet-20240620-v1:0": 200000,
|
||||
"us.anthropic.claude-3-5-haiku-20241022-v1:0": 200000,
|
||||
"us.anthropic.claude-3-5-sonnet-20241022-v2:0": 200000,
|
||||
@@ -263,11 +271,20 @@ LLM_CONTEXT_WINDOW_SIZES: Final[dict[str, int]] = {
|
||||
"us.meta.llama3-1-70b-instruct-v1:0": 128000,
|
||||
"us.meta.llama3-3-70b-instruct-v1:0": 128000,
|
||||
"us.meta.llama3-1-405b-instruct-v1:0": 128000,
|
||||
# Claude 4.x (eu-prefixed)
|
||||
"eu.anthropic.claude-sonnet-4-5-20250929-v1:0": 200000,
|
||||
"eu.anthropic.claude-sonnet-4-20250514-v1:0": 200000,
|
||||
"eu.anthropic.claude-opus-4-20250514-v1:0": 200000,
|
||||
# Claude 3.x (eu-prefixed)
|
||||
"eu.anthropic.claude-3-5-sonnet-20240620-v1:0": 200000,
|
||||
"eu.anthropic.claude-3-sonnet-20240229-v1:0": 200000,
|
||||
"eu.anthropic.claude-3-haiku-20240307-v1:0": 200000,
|
||||
"eu.meta.llama3-2-3b-instruct-v1:0": 131000,
|
||||
"eu.meta.llama3-2-1b-instruct-v1:0": 131000,
|
||||
# Claude 4.x (apac-prefixed)
|
||||
"apac.anthropic.claude-sonnet-4-5-20250929-v1:0": 200000,
|
||||
"apac.anthropic.claude-sonnet-4-20250514-v1:0": 200000,
|
||||
# Claude 3.x (apac-prefixed)
|
||||
"apac.anthropic.claude-3-5-sonnet-20240620-v1:0": 200000,
|
||||
"apac.anthropic.claude-3-5-sonnet-20241022-v2:0": 200000,
|
||||
"apac.anthropic.claude-3-sonnet-20240229-v1:0": 200000,
|
||||
@@ -275,6 +292,14 @@ LLM_CONTEXT_WINDOW_SIZES: Final[dict[str, int]] = {
|
||||
"amazon.nova-pro-v1:0": 300000,
|
||||
"amazon.nova-micro-v1:0": 128000,
|
||||
"amazon.nova-lite-v1:0": 300000,
|
||||
# Claude 4.x (no region prefix)
|
||||
"anthropic.claude-sonnet-4-5-20250929-v1:0": 200000,
|
||||
"anthropic.claude-sonnet-4-20250514-v1:0": 200000,
|
||||
"anthropic.claude-opus-4-5-20251101-v1:0": 200000,
|
||||
"anthropic.claude-opus-4-1-20250805-v1:0": 200000,
|
||||
"anthropic.claude-opus-4-20250514-v1:0": 200000,
|
||||
"anthropic.claude-haiku-4-5-20251001-v1:0": 200000,
|
||||
# Claude 3.x (no region prefix)
|
||||
"anthropic.claude-3-5-sonnet-20240620-v1:0": 200000,
|
||||
"anthropic.claude-3-5-haiku-20241022-v1:0": 200000,
|
||||
"anthropic.claude-3-5-sonnet-20241022-v2:0": 200000,
|
||||
|
||||
@@ -2038,18 +2038,47 @@ class BedrockCompletion(BaseLLM):
|
||||
"""Check if the model supports stop words."""
|
||||
return True
|
||||
|
||||
@staticmethod
|
||||
def _strip_region_prefix(model: str) -> str:
|
||||
"""Strip AWS region prefix from model ID.
|
||||
|
||||
Bedrock model IDs can be prefixed with region qualifiers like
|
||||
``us.``, ``eu.``, ``apac.``, or ``global.`` for cross-region
|
||||
inference. This helper returns the base model ID so that
|
||||
capability look-ups work regardless of the prefix.
|
||||
|
||||
Args:
|
||||
model: Full model ID, e.g. ``us.anthropic.claude-sonnet-4-20250514-v1:0``
|
||||
|
||||
Returns:
|
||||
Base model ID, e.g. ``anthropic.claude-sonnet-4-20250514-v1:0``
|
||||
"""
|
||||
region_prefixes = ("us.", "eu.", "apac.", "global.")
|
||||
for prefix in region_prefixes:
|
||||
if model.startswith(prefix):
|
||||
return model[len(prefix):]
|
||||
return model
|
||||
|
||||
def get_context_window_size(self) -> int:
|
||||
"""Get the context window size for the model."""
|
||||
from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO
|
||||
|
||||
# Context window sizes for common Bedrock models
|
||||
context_windows = {
|
||||
# Claude 4.x models
|
||||
"anthropic.claude-sonnet-4-5": 200000,
|
||||
"anthropic.claude-sonnet-4": 200000,
|
||||
"anthropic.claude-opus-4-5": 200000,
|
||||
"anthropic.claude-opus-4-1": 200000,
|
||||
"anthropic.claude-opus-4": 200000,
|
||||
"anthropic.claude-haiku-4-5": 200000,
|
||||
# Claude 3.x models
|
||||
"anthropic.claude-3-7-sonnet": 200000,
|
||||
"anthropic.claude-3-5-sonnet": 200000,
|
||||
"anthropic.claude-3-5-haiku": 200000,
|
||||
"anthropic.claude-3-opus": 200000,
|
||||
"anthropic.claude-3-sonnet": 200000,
|
||||
"anthropic.claude-3-haiku": 200000,
|
||||
"anthropic.claude-3-7-sonnet": 200000,
|
||||
"anthropic.claude-v2": 100000,
|
||||
"amazon.titan-text-express": 8000,
|
||||
"ai21.j2-ultra": 8192,
|
||||
@@ -2060,9 +2089,12 @@ class BedrockCompletion(BaseLLM):
|
||||
"deepseek.r1": 32768,
|
||||
}
|
||||
|
||||
# Strip region prefix (us., eu., apac., global.) for matching
|
||||
base_model = self._strip_region_prefix(self.model)
|
||||
|
||||
# Find the best match for the model name
|
||||
for model_prefix, size in context_windows.items():
|
||||
if self.model.startswith(model_prefix):
|
||||
if base_model.startswith(model_prefix):
|
||||
return int(size * CONTEXT_WINDOW_USAGE_RATIO)
|
||||
|
||||
# Default context window size
|
||||
@@ -2071,22 +2103,26 @@ class BedrockCompletion(BaseLLM):
|
||||
def supports_multimodal(self) -> bool:
|
||||
"""Check if the model supports multimodal inputs.
|
||||
|
||||
Claude 3+ and Nova Lite/Pro/Premier on Bedrock support vision.
|
||||
Claude 3+, Claude 4.x, and Nova Lite/Pro/Premier on Bedrock support vision.
|
||||
|
||||
Returns:
|
||||
True if the model supports images.
|
||||
"""
|
||||
model_lower = self.model.lower()
|
||||
vision_models = (
|
||||
# Strip region prefix so we only need base-model prefixes
|
||||
base_model = self._strip_region_prefix(self.model).lower()
|
||||
vision_prefixes = (
|
||||
# Claude 4.x models
|
||||
"anthropic.claude-sonnet-4",
|
||||
"anthropic.claude-opus-4",
|
||||
"anthropic.claude-haiku-4",
|
||||
# Claude 3.x models
|
||||
"anthropic.claude-3",
|
||||
# Amazon Nova models
|
||||
"amazon.nova-lite",
|
||||
"amazon.nova-pro",
|
||||
"amazon.nova-premier",
|
||||
"us.amazon.nova-lite",
|
||||
"us.amazon.nova-pro",
|
||||
"us.amazon.nova-premier",
|
||||
)
|
||||
return any(model_lower.startswith(m) for m in vision_models)
|
||||
return any(base_model.startswith(m) for m in vision_prefixes)
|
||||
|
||||
def _is_nova_model(self) -> bool:
|
||||
"""Check if the model is an Amazon Nova model.
|
||||
@@ -2096,8 +2132,8 @@ class BedrockCompletion(BaseLLM):
|
||||
Returns:
|
||||
True if the model is a Nova model.
|
||||
"""
|
||||
model_lower = self.model.lower()
|
||||
return "amazon.nova-" in model_lower
|
||||
base_model = self._strip_region_prefix(self.model).lower()
|
||||
return base_model.startswith("amazon.nova-")
|
||||
|
||||
def get_file_uploader(self) -> Any:
|
||||
"""Get a Bedrock S3 file uploader using this LLM's AWS credentials.
|
||||
|
||||
@@ -886,6 +886,225 @@ def test_bedrock_stop_sequences_sent_to_api():
|
||||
assert call_kwargs["inferenceConfig"]["stopSequences"] == ["\nObservation:", "\nThought:"]
|
||||
|
||||
|
||||
# =============================================================================
|
||||
# Claude V4 Model Support Tests
|
||||
# =============================================================================
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_model_detection():
|
||||
"""Test that Claude V4 model IDs are properly detected and routed to BedrockCompletion."""
|
||||
from crewai.llms.providers.bedrock.completion import BedrockCompletion
|
||||
|
||||
v4_models = [
|
||||
"bedrock/anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-1-20250805-v1:0",
|
||||
"bedrock/anthropic.claude-haiku-4-5-20251001-v1:0",
|
||||
"bedrock/anthropic.claude-sonnet-4-5-20250929-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-5-20251101-v1:0",
|
||||
]
|
||||
|
||||
for model_name in v4_models:
|
||||
llm = LLM(model=model_name)
|
||||
assert isinstance(llm, BedrockCompletion), f"Failed for model: {model_name}"
|
||||
assert llm.is_claude_model, f"is_claude_model should be True for {model_name}"
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_context_window_size():
|
||||
"""Test that Claude V4 models return correct context window sizes (200K)."""
|
||||
v4_models = [
|
||||
"bedrock/anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-1-20250805-v1:0",
|
||||
"bedrock/anthropic.claude-haiku-4-5-20251001-v1:0",
|
||||
"bedrock/anthropic.claude-sonnet-4-5-20250929-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-5-20251101-v1:0",
|
||||
]
|
||||
|
||||
for model_name in v4_models:
|
||||
llm = LLM(model=model_name)
|
||||
context_size = llm.get_context_window_size()
|
||||
# 200000 * 0.85 = 170000
|
||||
assert context_size > 150000, (
|
||||
f"Context window for {model_name} should be ~170K (200K * 0.85), got {context_size}"
|
||||
)
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_multimodal_support():
|
||||
"""Test that Claude V4 models are correctly detected as supporting multimodal."""
|
||||
v4_models = [
|
||||
"bedrock/anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-1-20250805-v1:0",
|
||||
"bedrock/anthropic.claude-haiku-4-5-20251001-v1:0",
|
||||
"bedrock/anthropic.claude-sonnet-4-5-20250929-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-5-20251101-v1:0",
|
||||
]
|
||||
|
||||
for model_name in v4_models:
|
||||
llm = LLM(model=model_name)
|
||||
assert llm.supports_multimodal(), (
|
||||
f"supports_multimodal() should return True for {model_name}"
|
||||
)
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_with_region_prefix():
|
||||
"""Test that Claude V4 models with region prefixes are properly handled."""
|
||||
from crewai.llms.providers.bedrock.completion import BedrockCompletion
|
||||
|
||||
region_prefixed_models = [
|
||||
"bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
"bedrock/eu.anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
"bedrock/apac.anthropic.claude-sonnet-4-5-20250929-v1:0",
|
||||
"bedrock/global.anthropic.claude-opus-4-20250514-v1:0",
|
||||
"bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0",
|
||||
]
|
||||
|
||||
for model_name in region_prefixed_models:
|
||||
llm = LLM(model=model_name)
|
||||
assert isinstance(llm, BedrockCompletion), f"Failed for model: {model_name}"
|
||||
|
||||
# Context window should still be correctly detected
|
||||
context_size = llm.get_context_window_size()
|
||||
assert context_size > 150000, (
|
||||
f"Context window for {model_name} should be ~170K, got {context_size}"
|
||||
)
|
||||
|
||||
# Multimodal should still be detected
|
||||
assert llm.supports_multimodal(), (
|
||||
f"supports_multimodal() should return True for {model_name}"
|
||||
)
|
||||
|
||||
|
||||
def test_bedrock_strip_region_prefix():
|
||||
"""Test that _strip_region_prefix correctly strips region qualifiers."""
|
||||
from crewai.llms.providers.bedrock.completion import BedrockCompletion
|
||||
|
||||
test_cases = [
|
||||
("us.anthropic.claude-sonnet-4-20250514-v1:0", "anthropic.claude-sonnet-4-20250514-v1:0"),
|
||||
("eu.anthropic.claude-3-5-sonnet-20241022-v2:0", "anthropic.claude-3-5-sonnet-20241022-v2:0"),
|
||||
("apac.anthropic.claude-sonnet-4-5-20250929-v1:0", "anthropic.claude-sonnet-4-5-20250929-v1:0"),
|
||||
("global.anthropic.claude-opus-4-20250514-v1:0", "anthropic.claude-opus-4-20250514-v1:0"),
|
||||
("anthropic.claude-sonnet-4-20250514-v1:0", "anthropic.claude-sonnet-4-20250514-v1:0"),
|
||||
("amazon.nova-pro-v1:0", "amazon.nova-pro-v1:0"),
|
||||
]
|
||||
|
||||
for input_model, expected in test_cases:
|
||||
result = BedrockCompletion._strip_region_prefix(input_model)
|
||||
assert result == expected, (
|
||||
f"_strip_region_prefix({input_model!r}) returned {result!r}, expected {expected!r}"
|
||||
)
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_inference_config():
|
||||
"""Test that inference config is properly prepared for Claude V4 models."""
|
||||
llm = LLM(
|
||||
model="bedrock/anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
temperature=0.7,
|
||||
top_p=0.9,
|
||||
top_k=40,
|
||||
max_tokens=2000,
|
||||
)
|
||||
|
||||
config = llm._get_inference_config()
|
||||
assert config["temperature"] == 0.7
|
||||
assert config["topP"] == 0.9
|
||||
assert config["topK"] == 40
|
||||
assert config["maxTokens"] == 2000
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_supports_function_calling():
|
||||
"""Test that Claude V4 models support function calling."""
|
||||
v4_models = [
|
||||
"bedrock/anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-opus-4-20250514-v1:0",
|
||||
"bedrock/anthropic.claude-sonnet-4-5-20250929-v1:0",
|
||||
]
|
||||
|
||||
for model_name in v4_models:
|
||||
llm = LLM(model=model_name)
|
||||
assert llm.supports_function_calling(), (
|
||||
f"supports_function_calling() should return True for {model_name}"
|
||||
)
|
||||
|
||||
|
||||
def test_bedrock_region_prefix_nova_model_detection():
|
||||
"""Test that _is_nova_model works with region-prefixed model IDs."""
|
||||
# Nova models with region prefix
|
||||
nova_llm = LLM(model="bedrock/us.amazon.nova-pro-v1:0")
|
||||
assert nova_llm._is_nova_model(), "us.amazon.nova-pro should be detected as Nova model"
|
||||
|
||||
# Claude model should not be detected as Nova
|
||||
claude_llm = LLM(model="bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0")
|
||||
assert not claude_llm._is_nova_model(), "Claude model should not be detected as Nova"
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_initialization_with_all_params():
|
||||
"""Test that Claude V4 models can be initialized with all supported parameters."""
|
||||
from crewai.llms.providers.bedrock.completion import BedrockCompletion
|
||||
|
||||
llm = LLM(
|
||||
model="bedrock/anthropic.claude-sonnet-4-20250514-v1:0",
|
||||
temperature=0.5,
|
||||
max_tokens=4096,
|
||||
top_p=0.95,
|
||||
top_k=50,
|
||||
stop_sequences=["\\nHuman:", "\\nAssistant:"],
|
||||
stream=True,
|
||||
region_name="us-west-2",
|
||||
)
|
||||
|
||||
assert isinstance(llm, BedrockCompletion)
|
||||
assert llm.model == "anthropic.claude-sonnet-4-20250514-v1:0"
|
||||
assert llm.temperature == 0.5
|
||||
assert llm.max_tokens == 4096
|
||||
assert llm.top_p == 0.95
|
||||
assert llm.top_k == 50
|
||||
assert llm.stream is True
|
||||
assert llm.region_name == "us-west-2"
|
||||
assert llm.is_claude_model is True
|
||||
|
||||
|
||||
def test_bedrock_claude_v4_call_with_mock(bedrock_mocks):
|
||||
"""Test that Claude V4 models can make API calls (mocked)."""
|
||||
_, mock_client = bedrock_mocks
|
||||
|
||||
mock_client.converse.return_value = {
|
||||
"output": {
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [{"text": "Hello from Claude Sonnet 4!"}],
|
||||
}
|
||||
},
|
||||
"usage": {"inputTokens": 10, "outputTokens": 8, "totalTokens": 18},
|
||||
}
|
||||
|
||||
llm = LLM(model="bedrock/anthropic.claude-sonnet-4-20250514-v1:0")
|
||||
result = llm.call("Hello")
|
||||
|
||||
assert result == "Hello from Claude Sonnet 4!"
|
||||
assert mock_client.converse.called
|
||||
|
||||
|
||||
def test_bedrock_v3_models_still_work():
|
||||
"""Regression test: ensure Claude V3 models still work after V4 changes."""
|
||||
from crewai.llms.providers.bedrock.completion import BedrockCompletion
|
||||
|
||||
v3_models = [
|
||||
"bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0",
|
||||
"bedrock/anthropic.claude-3-haiku-20240307-v1:0",
|
||||
"bedrock/anthropic.claude-3-opus-20240229-v1:0",
|
||||
"bedrock/anthropic.claude-3-7-sonnet-20250219-v1:0",
|
||||
]
|
||||
|
||||
for model_name in v3_models:
|
||||
llm = LLM(model=model_name)
|
||||
assert isinstance(llm, BedrockCompletion), f"Failed for model: {model_name}"
|
||||
assert llm.is_claude_model is True
|
||||
assert llm.get_context_window_size() > 150000
|
||||
assert llm.supports_multimodal()
|
||||
|
||||
|
||||
# =============================================================================
|
||||
# Agent Kickoff Structured Output Tests
|
||||
# =============================================================================
|
||||
|
||||
Reference in New Issue
Block a user