BerriAI/litellm · error · ContextWindowExceededError

AlephAlphaException

Error message

AlephAlphaException - {original_exception.message}

What it means

When Aleph Alpha's error text contains "This is longer than the model's maximum context length", litellm raises ContextWindowExceededError (BadRequestError subclass). Your prompt alone exceeded the served model's context — this mapping fires on the error string, before any status-code handling.

Solutions

1. Trim or summarize the prompt below the model's context limit
2. Pre-flight with litellm.token_counter(model, messages) compared to litellm.get_max_tokens(model)
3. Keep a sliding window over conversation history instead of replaying it fully
4. Move to a provider/model with a larger context if prompts are legitimately big

Example fix

# before
litellm.completion(model='aleph_alpha/luminous-supreme', messages=full_history)

# after
limit = litellm.get_max_tokens('aleph_alpha/luminous-supreme')
while litellm.token_counter(model='aleph_alpha/luminous-supreme', messages=full_history) > limit - 256:
    full_history.pop(1)  # drop oldest turns, keep system message
litellm.completion(model='aleph_alpha/luminous-supreme', messages=full_history)

model = 'aleph_alpha/luminous-base'
limit = litellm.get_max_tokens(model)
if litellm.token_counter(model=model, messages=msgs) >= limit - 256:
    msgs = trim_history(msgs, budget=limit - 256)  # keep system message, drop oldest

def is_context_window_error(e: BaseException) -> bool:
    return isinstance(e, litellm.ContextWindowExceededError)

try:
    resp = litellm.completion(model=model, messages=msgs)
except litellm.ContextWindowExceededError:
    resp = litellm.completion(model=model, messages=summarize_then_trim(msgs))

When it happens

Trigger: completion calls on model='aleph_alpha/<luminous-model>' where the prompt tokens exceed the model's context limit: whole documents pasted as input, accumulated conversation history, or large few-shot prefixes.

Common situations: Long chat histories never truncated; RAG retrieval returning more chunks than the model holds; test prompts sized for larger-context models reused on luminous-base.

Related errors

• AI21Exception
• Exception: Context Window Error
• ContextWindowExceededError
• TogetherAIException
• BedrockException: Context Window Error

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/82db2312393f3d7d. Report an issue: GitHub.

Appendix: source

Thrown at litellm/litellm_core_utils/exception_mapping_utils.py:1748

            message=f"TogetherAIException - {original_exception.message}",
            llm_provider="together_ai",
            model=model,
            request=getattr(original_exception, "request", None),
        )


def _map_aleph_alpha_exception(
    *,
    model: str,
    original_exception: _ProviderHTTPException,
    custom_llm_provider: str,
    error_str: str,
    exception_type: str,
    exception_provider: str,
    extra_information: str,
) -> None:
    if "This is longer than the model's maximum context length" in error_str:
        raise ContextWindowExceededError(
            message=f"AlephAlphaException - {original_exception.message}",
            llm_provider="aleph_alpha",
            model=model,
            response=getattr(original_exception, "response", None),
        )
    elif "InvalidToken" in error_str or "No token provided" in error_str:
        raise BadRequestError(
            message=f"AlephAlphaException - {original_exception.message}",
            llm_provider="aleph_alpha",
            model=model,
            response=getattr(original_exception, "response", None),
        )
    elif hasattr(original_exception, "status_code"):
        verbose_logger.debug("status code: %s", original_exception.status_code)
        if original_exception.status_code == 401:
            raise AuthenticationError(
                message=f"AlephAlphaException - {original_exception.message}",
                llm_provider="aleph_alpha",

View on GitHub (pinned to 77b7c6c40c)