API Benchmark Scores Do Not Reliably Transfer to Chatbot Interfaces
Read the original on arXiv AI →The Flow has not summarised this story yet — read it at arXiv AI.
The Flow has not summarised this story yet — read it at arXiv AI.
The paper examines whether benchmark scores obtained via APIs accurately reflect the performance of AI chatbots when accessed through user interfaces. By auditing ChatGPT, Claude, and Gemini across seven systems and nine benchmarks, the authors find that API evaluations consistently overestimate accuracy and consistency compared to interface evaluations, with differences comparable to downgrading a full model generation. Attempts to align API behavior with interface behavior through prompt and parameter adjustments only partially close the gap.
arXiv:2608. 06202v1 Announce Type: cross Abstract: Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readiness.
The paper evaluates 20 third‑party API relays that provide access to large language models such as GPT‑6 Astra and Claude Opus 5. It finds that many low‑price relays exhibit answer‑behavior differences, some return tool calls even when disabled, and several report inflated input‑token counts leading to higher charges than the official API. The study demonstrates that advertised models and prices alone do not guarantee reliable service, highlighting the need for repeated, detailed evaluations.
The paper evaluates 20 third‑party LLM API relays that claim to provide GPT‑6 Astra and Claude Opus 5 at lower prices. By comparing answers to official responses, inspecting tool permissions, and reconciling billed token counts, the study finds that many relays exhibit behavioral differences, return unexpected tool calls, and report inflated input‑token usage. These discrepancies highlight that advertised models and prices alone do not guarantee reliable service.
ADeptS-Bench is a new benchmark designed to assess the trustworthiness of Computer Use Agents (CUAs) across mobile and desktop devices. It consists of two streams: a Safety stream with paired benign and malicious tasks that embed visual threats, and a Disambiguation stream that tests whether agents seek clarification when instructions are ambiguous. Evaluation of seven models shows none consistently achieves high task success while keeping attack success low, and all models exhibit problematic behaviors such as unhesitant checkout on a $25K order and failure to detect a mislabeled factory reset button.
arXiv:2607. 02587v2 Announce Type: replace-cross Abstract: Trust-benchmark scores reported on a chat-LLM release line are often carried across several checkpoints of the same line, as if the underlying model had not shifted between releases.