xtekky/gpt4free · error · RateLimitError

{message}: {model}

Error message

{message}: {model}

What it means

Raised by raise_gemini_error() after mapping a Gemini error code (extracted from the response) through GEMINI_ERROR_MESSAGES. Codes 1037 and 1060 raise RateLimitError; every other code raises ResponseError. The message combines the human-readable code description with the model name, so the caller knows which model and which failure class is involved.

Source

Thrown at g4f/Provider/needs_auth/gemini_utils.py:265

    if not isinstance(value, list):
        return None
    if value and value[0] == "wrb.fr":
        code = get_nested_value(value, [5, 2, 0, 1, 0])
        if isinstance(code, int):
            return code
    for item in value:
        code = extract_gemini_error_code(item)
        if code is not None:
            return code
    return None


def raise_gemini_error(code: int, model: str) -> None:
    message = GEMINI_ERROR_MESSAGES.get(
        code, f"Gemini rejected the request with error code {code}"
    )
    if code in (1037, 1060):
        raise RateLimitError(f"{message}: {model}")
    raise ResponseError(f"{message}: {model}")

View on GitHub (pinned to 973504e177)

Solutions

1. Match the message text against the code table in GEMINI_ERROR_MESSAGES to identify the exact code
2. For 1037/1060 (RateLimitError): back off and retry after the quota window resets
3. For other codes: fix the underlying cause (prompt content for safety codes, model name for not-found codes)
4. Switch model or provider if the code indicates a permanent rejection

from g4f.errors import RateLimitError, ResponseError
try:
    result = await provider.create_async(model, messages)
except RateLimitError as e:
    # codes 1037/1060: quota window
    await asyncio.sleep(60)
    result = await provider.create_async(model, messages)
except ResponseError as e:
    # non-rate code: inspect message, do not blind-retry
    raise

When it happens

Trigger: Calling a Gemini-based provider and receiving a response containing a structured error code (via extract_gemini_error_code) — e.g. quota/recursion (1037, 1060) or safety/policy rejection codes — triggering raise_gemini_error.

Common situations: Hitting Gemini free-tier quota limits, safety filters rejecting the prompt, model deprecation, or invalid model names routed through the Gemini API surface.

Related errors

• Failed to chat: {response.status} {error_text}
• Failed to decode JSON from PhindAi response: {text}
• PhindAi API returned success=False: {data}
• {data['code']}:{data['details']}
• Response: {resp_json}

AI-assisted analysis of xtekky/gpt4free@973504e177 (2026-08-14). Data as JSON: /api/errors/5b26f93edfebb6d5. Report an issue: GitHub.