static audit · 2026-10-25T20:30:57.018Z

openai/openai-node

120 source files scanned · 36 LLM call site(s) · ~0% of LLM spend looks recoverable (static heuristics · repo looks like a library)

Library/SDK layout detected (package "openai" is an LLM SDK/framework): call sites are most likely pass-through wrappers where the caller chooses the model and prompt, so findings are downgraded one level and the estimate is discounted.

$80per 10k runs today (gpt-4o, ~2000 prompt tokens)

$80per 10k runs with fixes applied

![TokenZero audit](https://tokenzero.sh/v1/badge/audit/aud_92JkzM4vXRXwLxHvkAbXSQ.svg)

[![LLM spend audited by TokenZero](https://tokenzero.sh/v1/badge/repo/openai/openai-node.svg)](https://tokenzero.sh/audit/openai/openai-node)

Models referenced

gpt-3.5-turbo ×9
gpt-4o-2024-08-06 ×7
gpt-4 ×6
gpt-5.5 ×2 (cheaper sibling: gpt-5.4-mini)
gpt-4o-mini ×2
gpt-5.6-sol ×1 (cheaper sibling: gpt-5.6-terra)

Findings

Severity	Where	Finding	Est.
info	`examples/chat-completions/tool-calls-stream.ts:121`	LLM call inside a loop Per-item calls repeat the full prompt prefix for every element. If items are independent and latency-tolerant, the provider Batch API is 50% cheaper; if the prefix is shared, caching pays off immediately. Fix: Batch where possible; otherwise keep the shared prefix cache-formatted (TokenZero does this automatically).	25%
info	`examples/chat-completions/function-call-stream-raw.ts:87`	LLM call inside a loop Per-item calls repeat the full prompt prefix for every element. If items are independent and latency-tolerant, the provider Batch API is 50% cheaper; if the prefix is shared, caching pays off immediately. Fix: Batch where possible; otherwise keep the shared prefix cache-formatted (TokenZero does this automatically).	25%
info	`examples/chat-completions/function-call-stream.ts:87`	LLM call inside a loop Per-item calls repeat the full prompt prefix for every element. If items are independent and latency-tolerant, the provider Batch API is 50% cheaper; if the prefix is shared, caching pays off immediately. Fix: Batch where possible; otherwise keep the shared prefix cache-formatted (TokenZero does this automatically).	25%
info	`examples/chat-completions/function-call-diy.ts:82`	LLM call inside a loop Per-item calls repeat the full prompt prefix for every element. If items are independent and latency-tolerant, the provider Batch API is 50% cheaper; if the prefix is shared, caching pays off immediately. Fix: Batch where possible; otherwise keep the shared prefix cache-formatted (TokenZero does this automatically).	25%
info	`examples/chat-completions/function-call.ts:82`	LLM call inside a loop Per-item calls repeat the full prompt prefix for every element. If items are independent and latency-tolerant, the provider Batch API is 50% cheaper; if the prefix is shared, caching pays off immediately. Fix: Batch where possible; otherwise keep the shared prefix cache-formatted (TokenZero does this automatically).	25%
info	`examples/chat-completions/tool-calls-stream.ts:121`	Unrecognised model id "gpt-3.5-turbo" Not in the current pricing table: it may be retired or renamed. Retired models return errors or are silently remapped by some providers. Fix: Check the provider model list; TokenZero flags unpriced models in `X-TokenZero-Warning`.	0%
info	`examples/chat-completions/chat-params-types.ts:16`	Unrecognised model id "gpt-4" Not in the current pricing table: it may be retired or renamed. Retired models return errors or are silently remapped by some providers. Fix: Check the provider model list; TokenZero flags unpriced models in `X-TokenZero-Warning`.	0%

Fix all of it with one line

OPENAI_BASE_URL=https://tokenzero.sh/v1
ANTHROPIC_BASE_URL=https://tokenzero.sh

Keys pass through untouched. Cache formatting and exact caching are on by default; cascade routing stays opt-in. How it works