Home Model series GPT nano GPT-5 nano GPT-5 nano · LongFact-Objects · GPT-5 developer release; no tools · Scores and configurations GPT-5 nano · LongFact-Objects · GPT-5 developer release; no tools · Scores and configurations GPT-5 nano record for LongFact-Objects · GPT-5 developer release; no tools. 3 score records for this release, including 3 exact values (may include historical records). Includes 3 source locations.
Scores and evaluation conditions GPT-5 nano · LongFact-Objects scores and configurations Metric and score Configuration and evaluation conditions Sources Hallucination rate 2.8% Reported 2.8% · Unit % Published 2025-08-07
GPT-5 nano (high) Settings API nano model; benchmark table labels high reasoning effort. Unknowns none recorded Tools Tools forbidden Reasoning budget high Framework not recorded Protocol LongFact-Objects hallucination rate (no tools; lower is better) Conditions API GPT-5 nano; high reasoning effort; no tools; lower is better. View this score in rankings Hallucination rate 1.9% Reported 1.9% · Unit % Published Unconfirmed
GPT-5 thinking · browsing on Settings System Card thinking model; browsing enabled; OpenAI o3 claim extraction and web-based factuality grading. Unknowns 图未注明具体reasoning effort、采样参数或重复次数。 Tools Tools allowed Reasoning budget not recorded Framework not recorded Protocol LongFact-Objects; thinking model; browsing on; claim-level hallucination rate; OpenAI o3 grader. Conditions Explicit Figure 2/3 numeric labels; effort not inferred from a different API announcement run; no curve estimation. View this score in rankings Hallucination rate 3.1% Reported 3.1% · Unit % Published Unconfirmed
GPT-5 thinking · browsing off Settings System Card thinking model; browsing disabled; OpenAI o3 claim extraction and web-based factuality grading. Unknowns 图未注明具体reasoning effort、采样参数或重复次数。 Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol LongFact-Objects; thinking model; browsing off; claim-level hallucination rate; OpenAI o3 grader. Conditions Explicit Figure 2/3 numeric labels; effort not inferred from a different API announcement run; no curve estimation. View this score in rankings
Benchmark versions and metrics Aliases LongFact Objects
Hallucination rate · % · Lower is better