OpenAI API Errors — Complete Fix Directory
Every OpenAI error, decoded and fixed. GPT-5, GPT-4o, o-series reasoning models, Assistants API. Tested against the live API, updated November 2026.
15
Error pages live
6
Categories covered
3
Code languages
Nov 2026
Last tested
gpt-5
gpt-5-mini
gpt-4o
gpt-4o-mini
o3
o3-mini
o1
text-embedding-3-large
R
Rate Limits & Context
TPM, RPM, TPD limits, context window overflows
429 rate_limit
Rate Limit Exceeded (429)
RPM/TPM caps hit. Backoff, tier limits, and preventing burst overshoots.
400 context_length
Context Length Exceeded
Prompt + max_tokens exceeds model window. Trimming, summarization, larger models.
429 TPM
Tokens Per Minute Limit Hit
Distinct from RPM. Batching, rate smoothing, tier upgrades.
F
Function Calling / Tools
JSON validation, tool_choice, function schemas
tool_call parse
Function Call Invalid JSON
Model returned malformed arguments. Structured Outputs, validation, retry patterns.
arguments parse
Function Arguments Parse Error
arguments is a string, not object. JSON.parse before use.
400 tool_choice
tool_choice Mismatch
Specified tool doesn't exist in tools array. All modes explained.
function_not_found
Function Name Not Found
Model called a tool that doesn't exist. Hallucination handling.
O
o-Series Reasoning Models
o3, o3-mini, o1 specific errors and quirks
reasoning_tokens
o-Series Reasoning Tokens Exhausted
max_completion_tokens too low for reasoning + output. Budget math.
400 reasoning_effort
reasoning_effort Invalid Value
low/medium/high only. Cost multipliers and task fit.
400 message role
developer vs system Role Conflict
o-series uses "developer" not "system". Cross-model compatibility.
M
Moderation & Safety
Input/output blocking, policy violations
S
Structured Outputs
JSON mode, response_format, schema validation
A
Assistants API
Threads, runs, and long-running workflows