2021-10 · GPT-3 early apps
GPT-3 early apps · landscape continues
GPT-3 early apps · Scores are era-relative (month top ≈95–100) for landscape review — not official historical Artificial Analysis replays. From 2026-08, the latest month aligns with the live AA-style snapshot.
Score method：Scores are era-relative (month top ≈95–100) for landscape review — not official historical Artificial Analysis replays. From 2026-08, the latest month aligns with the live AA-style snapshot.
Month landscape board
| # | Model | Vendor | Era score | Elo≈ | Pricing | Highlight |
|---|---|---|---|---|---|---|
| 🥇 | GPT-3 davinci | OpenAI | 92 | 1162 | 按 token 计费，远高于后来 | API 写作/补全主流 |
| 🥈 | Jurassic-1 | AI21 | 78 | 1150 | — | 早期竞品 API |
| 🥉 | Megatron-Turing | NVIDIA/MS | 74 | 1146 | — | 研究向 |
Elo≈ is a LMSYS-style discussion-scale estimate for recap — not an official historical snapshot.
Launches this month
No separately tagged blockbuster launches this month (landscape continues).
Pricing changes
No separately tagged pricing events (prior month scale carries over).
Benchmark context
Benchmark context: public discussion of Arena Elo, AA Intelligence, and open suites (GPQA / SWE-bench, etc.). This page is a narrative archive, not a live probe.
Some historical narratives remain bilingual; model names stay in original form.
Related tools
Clavue CLI / imux IDE / clavue-2.1
Clavue · CLI 与 Agent 平台
CLI / GUI Agent 运行时、设备登录、会员额度与 OpenAI 兼容 API（api.clavue.com）。开发脚本、CI 与本地工具一条链路。
打开 Clavue ↗imux · 原生 macOS AI IDE
Clavue 平台的原生 IDE：多 Agent 分屏、Ghostty 级终端、Agent Chat、浏览器自动化与 Supervisor——不是又一个 Electron 壳。
了解 imux ↗Clavue 2.1 · 产品级大模型
旗舰模型 clavue-2.1（及 fast / pro / rev）：适合复杂推理与长程 Agent 循环。Chat、imux、API 共用会员额度。
查看模型 ↗