agno-agi/agno
120 source files scanned · 22 LLM call site(s) · ~1% of LLM spend looks recoverable (static heuristics)
$70per 10k runs today (claude-sonnet-5-5, ~2000 prompt tokens)
$69.3per 10k runs with fixes applied
[](https://tokenzero.sh/audit/agno-agi/agno)
Models referenced
claude-sonnet-5-5 ×8 (cheaper sibling: claude-haiku-5-5)gpt-5.4-mini ×4gemini-3.7-flash ×2
Findings
| Severity | Where | Finding | Est. |
|---|---|---|---|
| medium | libs/agno/agno/models/anthropic/claude.py:841 | 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, and OpenAI Flex processing (`service_tier: "flex"`) gives the same 50% on ordinary calls without restructuring; if the prefix is shared, caching pays off immediately. Fix: Batch where possible, or set `service_tier: "flex"` on OpenAI calls that can wait; otherwise keep the shared prefix cache-formatted (TokenZero does this automatically). | 25% |
| info | libs/agno/agno/models/google/gemini.py:567 | Unrecognised model id "gemini-3.7-flash" 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