Topic hub pagination
Continue browsing this topic cluster with SEO-safe static pagination.
Fix LiteLLM embedding router not retrying or failing over when embedding host is unreachable Includes evidence for LiteLLM troubleshooting demand.
aembedding() missing num_retries kwarg — no failover for embedding model groups Fix LiteLLM OCI adapter incorrectly mapping max_completion_tokens to maxTokens for GPT-5 models Includes evidence for LiteLLM troubleshooting demand.
Unsupported parameter: 'max_tokens' is not supported Fix LiteLLM proxy failover not working when a deployment host is unreachable Includes evidence for LiteLLM troubleshooting demand.
APIConnectionError Fix LiteLLM daily activity API timezone counting bug overstates totals Includes evidence for LiteLLM troubleshooting demand.
LiteLLM /user/daily/activity overstates totals for single-day window in non-UTC timezones Fix LiteLLM Router.aspeech no retry failover for TTS requests Includes evidence for LiteLLM troubleshooting demand.
LiteLLM Router.aspeech() bypasses async_function_with_fallbacks — TTS requests have no retry or failover Fix LiteLLM proxy TPM limit not being enforced correctly across multiple replicas in production deployments Includes evidence for LiteLLM troubleshooting demand.
tpm_limit × N_replica — TPM enforcement per-pod instead of cross-pod in multi-replica deployments Check if LiteLLM package is safe / recover from compromised PyPI install Includes evidence for LiteLLM troubleshooting demand.
litellm PyPI package (v1.82.7 + v1.82.8) compromised — credential theft via malicious code in proxy_server.py and litellm_init.pth Fix LiteLLM llm_as_a_judge guardrail failing with BadRequestError when using vLLM self-hosted model Includes evidence for LiteLLM troubleshooting demand.
litellm.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. Fix missing x-ratelimit headers on LiteLLM proxy streaming responses Includes evidence for LiteLLM troubleshooting demand.
x-ratelimit-* headers dropped on streaming and plain-dict responses (v3 parallel_request_limiter) Fix LiteLLM returning 400 errors when using OpenAI Codex CLI with Chinese domestic models via protocol translation Includes evidence for LiteLLM troubleshooting demand.
400 Bad Request — LiteLLM passes unsupported parameters to Chinese models for Codex CLI Fix 400 error when using LiteLLM with Codex CLI and Chinese/domestic LLM models Includes evidence for LiteLLM troubleshooting demand.
400 Bad Request — unsupported parameter passed to model API Fix LiteLLM ChatGPT image generation returning 403 Cloudflare challenge page Includes evidence for LiteLLM troubleshooting demand.
Enable JavaScript and cookies to continue Fix LiteLLM image generation failing with Cloudflare 403 challenge for ChatGPT subscriptions Includes evidence for LiteLLM troubleshooting demand.
Enable JavaScript and cookies to continue — 403 Cloudflare challenge on image generation Fix LiteLLM proxy database connection failure on Windows Includes evidence for LiteLLM troubleshooting demand.
Database connection failed when running in database mode on Windows Fix LiteLLM spend_log_cleanup silent failure on Kubernetes Includes evidence for LiteLLM troubleshooting demand.
spend_log_cleanup.py:153 - Error during cleanup Fix LiteLLM virtual key incorrectly rejecting requests with BudgetExceededError despite /key/info showing spend below max_budget Includes evidence for LiteLLM troubleshooting demand.
BudgetExceededError: spend exceeds max_budget for virtual key Fix missing rate limit headers on LiteLLM streaming API responses so clients can properly throttle requests Includes evidence for LiteLLM troubleshooting demand.
x-ratelimit-remaining / x-ratelimit-limit headers missing on streaming responses Fix LiteLLM deployment-level TPM rate limit being enforced per-pod instead of across all replicas in multi-replica Kubernetes deployments Includes evidence for LiteLLM troubleshooting demand.
Deployment-level TPM limit enforced per-pod, not cross-pod — effective limit = tpm_limit × N_replicas Fix LiteLLM crash with TypeError 'async for requires an object with aiter method' when streaming from models with reasoning/thinking output Includes evidence for LiteLLM troubleshooting demand.
TypeError: 'async for' requires an object with __aiter__ method, got NoneType Fix LiteLLM BudgetExceededError rejecting requests when actual spend is below budget limit Includes evidence for LiteLLM troubleshooting demand.
BudgetExceededError: stale spend reported while /key/info shows spend below max_budget