BerriAI/litellm · error · ContextWindowExceededError

BedrockException: Context Window Error - {error_str}

Error message

BedrockException: Context Window Error - {error_str}

What it means

Bedrock-specific mapping: when the AWS error text matches 'too many tokens', 'expected maxLength:', 'Input is too long', 'prompt is too long', 'prompt: length: 1..', or 'Too many input tokens', LiteLLM raises ContextWindowExceededError with llm_provider='bedrock'. The prompt exceeds the model's context window as enforced by Bedrock.

Source

Thrown at litellm/litellm_core_utils/exception_mapping_utils.py:832

def _map_bedrock_exception(
    *,
    model: str,
    original_exception: _ProviderHTTPException,
    custom_llm_provider: str,
    error_str: str,
    exception_type: str,
    exception_provider: str,
    extra_information: str,
) -> None:
    if (
        "too many tokens" in error_str
        or "expected maxLength:" in error_str
        or "Input is too long" in error_str
        or "prompt is too long" in error_str
        or "prompt: length: 1.." in error_str
        or "Too many input tokens" in error_str
    ):
        raise ContextWindowExceededError(
            message=f"BedrockException: Context Window Error - {error_str}",
            model=model,
            llm_provider="bedrock",
        )
    elif "Conversation blocks and tool result blocks cannot be provided in the same turn." in error_str:
        raise BadRequestError(
            message=f"BedrockException - {error_str}\n. Enable 'litellm.modify_params=True' (for PROXY do: `litellm_settings::modify_params: True`) to insert a dummy assistant message and fix this error.",
            model=model,
            llm_provider="bedrock",
            response=getattr(original_exception, "response", None),
        )
    elif "Malformed input request" in error_str:
        raise BadRequestError(
            message=f"BedrockException - {error_str}",
            model=model,
            llm_provider="bedrock",
            response=getattr(original_exception, "response", None),
        )

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Trim/summarize history and retrieved chunks before the call.
2. Check the limit with litellm.get_max_tokens(model='bedrock/<model-id>') and count with litellm.token_counter.
3. Switch to a Bedrock model with a larger context (e.g. Claude 3.5+ with 200k).
4. Reduce tool definitions or compress the system prompt.

Example fix

# before
resp = litellm.completion(model="bedrock/amazon.titan-text-express-v1", messages=msgs)

# after
model = "bedrock/anthropic.claude-3-5-sonnet-20240620-v1:0"
max_tok = litellm.get_max_tokens(model)
while litellm.token_counter(model=model, messages=msgs) > max_tok - 512:
    msgs.pop(1)
resp = litellm.completion(model=model, messages=msgs)

import litellm

def fits_bedrock_context(bedrock_model: str, messages: list, reserve: int = 256) -> bool:
    limit = litellm.get_max_tokens(model=bedrock_model) or 4096
    return litellm.token_counter(model=bedrock_model, messages=messages) <= limit - reserve

import litellm

def is_bedrock_context_error(e: BaseException) -> bool:
    return isinstance(e, litellm.ContextWindowExceededError) and getattr(e, "llm_provider", "") == "bedrock"

try:
    resp = litellm.completion(model="bedrock/...", messages=messages)
except litellm.ContextWindowExceededError:
    messages = summarize_history(messages)
    resp = litellm.completion(model="bedrock/<larger-context-model>", messages=messages)

When it happens

Trigger: bedrock completion/converse calls where input tokens exceed the model limit — e.g. Titan (4k/8k), older Claude on Bedrock, Llama models with small windows, or huge tool definitions plus long history.

Common situations: Assuming a large context because the API accepts it, but the Bedrock model id has a tighter limit; RAG over-stuffing; or summing system prompt + tools + history beyond e.g. 4096 tokens on small-window models.

Related errors

• {custom_llm_provider.capitalize()}Exception: Context Window
• BedrockException Invalid Authentication - {error_str}
• BedrockException PermissionDeniedError - {error_str}
• BedrockException: Rate Limit Error - {error_str}
• BedrockException: Timeout Error - {error_str}

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/395e524d24fb1ff3. Report an issue: GitHub.