GPT nano · 发布历史
OpenAI GPT nano 系列已收录 2 次发布和 39 个测试版本。
GPT nano 轻量分支,独立记录各代型号与官方前代关系。
全部版本
- GPT-5.4 nano14 项测试
- GPT-5 nano30 项测试
已收录测试 · 39
- SWE-Bench Pro · PublicGPT-5.4 nano
- Terminal-Bench · 2.0;代理框架按协议区分GPT-5.4 nano
- Toolathlon · Release article does not state benchmark version or scoring protocol (2026-03)GPT-5.4 nano
- GPQA Diamond · 198-question Diamond subsetGPT-5.4 nano、GPT-5 nano
- Humanity's Last Exam · 2,500-question benchmarkGPT-5.4 nano、GPT-5 nano
- OSWorld-Verified · Verified benchmark; evaluation protocol noted per sourceGPT-5.4 nano
- MCP-Atlas · 所选材料未明确子集GPT-5.4 nano
- τ²-bench · Telecom sectionGPT-5.4 nano、GPT-5 nano
- MMMU-Pro · Standard/Vision 等格式未在所选材料中注明GPT-5.4 nano
- OmniDocBench 1.5 · Version 1.5; average edit distance across text, formula, table and reading orderGPT-5.4 nano
- MRCR-V2 · 8-needle; up to 128K average contextGPT-5.4 nano
- OpenAI MRCR · v2, 256K context, 8 needlesGPT-5.4 nano
- GraphWalks · BFS;0–128K 上下文;题数未注明GPT-5.4 nano、GPT-5 nano
- GraphWalks · parents;0–128K 上下文;题数未注明GPT-5.4 nano、GPT-5 nano
- AIME · 2025 editionGPT-5 nano
- SWE-bench Verified · GPT-5 发布评测固定子集;500 题中排除 23 题,n=477GPT-5 nano
- MMMU · Single-attemptGPT-5 nano
- FrontierMath · Set/version unspecified in GPT-5 and GPT-5.1 release tables; Python tool enabled.GPT-5 nano
- HMMT 2025 · 2025 edition; no-tools protocol from GPT-5 developer releaseGPT-5 nano
- MMMU-Pro · Standard and Vision formatsGPT-5 nano
- CharXiv Reasoning · Version/split not stated in selected sourceGPT-5 nano
- Video-MMMU · Test subsetGPT-5 nano
- ERQA · Version unspecified in GPT-5 developer releaseGPT-5 nano
- SWE-Lancer · IC Diamond setGPT-5 nano
- Aider Polyglot · Diff or diff-fenced edit formatGPT-5 nano
- Scale MultiChallenge · o3-mini grader protocol from GPT-5 developer releaseGPT-5 nano
- Internal API instruction-following evaluation · Hard subset; exact set/version unspecifiedGPT-5 nano
- COLLIE · Version unspecified in GPT-5 developer releaseGPT-5 nano
- tau2-bench · Airline categoryGPT-5 nano
- τ²-bench · Retail sectionGPT-5 nano
- OpenAI MRCR · 2 needles; 128k contextGPT-5 nano
- OpenAI MRCR · 2 needles; 256k contextGPT-5 nano
- BrowseComp Long Context · 128k contextGPT-5 nano
- BrowseComp Long Context · 256k contextGPT-5 nano
- VideoMME · Long subset, with subtitle categoryGPT-5 nano
- LongFact-Concepts · GPT-5 developer release; no toolsGPT-5 nano
- LongFact-Objects · GPT-5 developer release; no toolsGPT-5 nano
- FActScore · GPT-5 developer release; no tools; hallucination-rate metricGPT-5 nano
- SimpleQA · No search-enabled; report F1GPT-5 nano