diff --git a/README.en.md b/README.en.md index a1643eb..f2ff6cb 100644 --- a/README.en.md +++ b/README.en.md @@ -34,6 +34,7 @@ One-shot LLM eval cases by NAGI STUDIO: same prompt, different model x harness x | Gemini 3.5 Flash | Google | AntiGravity · High
Google AI Studio · High | 03 | | Grok Build | xAI | Grok Build TUI · Max | 02 | | Composer 2.5 | Cursor | Cursor · Max
Cursor · Default
Grok Build TUI · Default | 05 | +| Mistral Medium 3.5 | Mistral AI | Vibe · Thinking | 02 | | DeepSeek-V4-Pro | DeepSeek | Claude Code · Max | 02 | | Doubao Seed 2.0 Pro | ByteDance | Doubao Web · Pro Mode | 01 | | Doubao Seed 2.0 Mini | ByteDance | Doubao Web · Fast Mode | 01 | @@ -41,7 +42,7 @@ One-shot LLM eval cases by NAGI STUDIO: same prompt, different model x harness x | Kimi K2.6 | Moonshot AI | Kimi Code · Thinking | 02 | | MiniMax M3 | MiniMax | MiniMax Code Web · Thinking | 01 | -Pending: Mistral Medium 3.5 · GLM-5.1 · Qwen3.7-Max (PRs welcome) +Pending: GLM-5.1 · Qwen3.7-Max (PRs welcome) > This table is a hand-maintained snapshot; the site's registry section is generated live from `models/*.json` and is authoritative. diff --git a/README.md b/README.md index bf71e27..471adbc 100644 --- a/README.md +++ b/README.md @@ -34,6 +34,7 @@ NAGI STUDIO 的 LLM 测评案例集:同一段提示词,不同「模型 × Ha | Gemini 3.5 Flash | Google | AntiGravity · High
Google AI Studio · High | 03 | | Grok Build | xAI | Grok Build TUI · Max | 02 | | Composer 2.5 | Cursor | Cursor · Max
Cursor · Default
Grok Build TUI · Default | 05 | +| Mistral Medium 3.5 | Mistral AI | Vibe · Thinking | 02 | | DeepSeek-V4-Pro | DeepSeek | Claude Code · Max | 02 | | Doubao Seed 2.0 Pro | ByteDance | Doubao Web · Pro Mode | 01 | | Doubao Seed 2.0 Mini | ByteDance | Doubao Web · Fast Mode | 01 | @@ -41,7 +42,7 @@ NAGI STUDIO 的 LLM 测评案例集:同一段提示词,不同「模型 × Ha | Kimi K2.6 | Moonshot AI | Kimi Code · Thinking | 02 | | MiniMax M3 | MiniMax | MiniMax Code Web · Thinking | 01 | -待测:Mistral Medium 3.5 · GLM-5.1 · Qwen3.7-Max(欢迎 PR 补充) +待测:GLM-5.1 · Qwen3.7-Max(欢迎 PR 补充) > 此表为人工维护的快照;站点上的「模型阵容」表由 `models/*.json` 实时生成,以站点为准。 diff --git a/models/mistral-medium-3-5.json b/models/mistral-medium-3-5.json index 4ad0737..6ee7699 100644 --- a/models/mistral-medium-3-5.json +++ b/models/mistral-medium-3-5.json @@ -1,5 +1,23 @@ { "label": "Mistral Medium 3.5", "vendor": "Mistral AI", - "order": 80 + "harness": "Vibe", + "effort": "Thinking", + "order": 80, + "runs": { + "mythos-craft": { + "note": { + "zh": "在 Vibe 网页版以 Mistral Medium 3.5(开启 Thinking)一次生成;本页为修复版(仅改 1 行)。问题:初始化时一次性同步生成 RENDER_DISTANCE=4 的 9×9=81 个区块,而每个区块要对 16×16×128 个体素逐一在 mythStructures 数组里做线性查找(再加 6 个相邻面),约 O(体素×建筑)≈ 数十亿次/区块。修复前:加载后主线程被卡死,页面一直停在加载界面、始终无法打开。原因:首屏同步生成量过大且单区块为暴力线性查找,首帧渲染前即阻塞主线程。将 RENDER_DISTANCE 由 4 改为 1(首屏改为同步生成 3×3=9 个区块)后即可正常打开并以 60fps 运行。地形、画面与玩法均为模型原始产出,未作他改。", + "en": "Generated in one shot in the Vibe web app on Mistral Medium 3.5 (Thinking enabled); fixed build (1 line changed). Bug: on init the world synchronously builds the full RENDER_DISTANCE=4 ring — a 9x9 = 81-chunk grid — and each chunk runs a linear scan over the mythStructures array for every one of its 16x16x128 voxels (plus six neighbor faces), roughly O(voxels x structures) ≈ billions of ops per chunk. Before the fix: the main thread froze on load and the page stayed on the loading screen, never opening. Cause: the first-screen synchronous generation is far too large and each chunk uses a brute-force linear lookup, blocking the main thread before the first frame. Setting RENDER_DISTANCE from 4 to 1 (a 3x3 = 9-chunk first-screen load) lets it open and run at 60fps. Terrain, visuals and gameplay are the model's original output, otherwise unmodified." + }, + "contributor": "Nagi-ovo" + }, + "pelican-cycling": { + "note": { + "zh": "在 Vibe 网页版以 Mistral Medium 3.5(开启 Thinking)一次生成,未作修改。", + "en": "Generated in one shot in the Vibe web app on Mistral Medium 3.5 (Thinking enabled), unmodified." + }, + "contributor": "Nagi-ovo" + } + } } diff --git a/outputs/mistral-medium-3-5/mythos-craft.html b/outputs/mistral-medium-3-5/mythos-craft.html new file mode 100644 index 0000000..7c8e552 --- /dev/null +++ b/outputs/mistral-medium-3-5/mythos-craft.html @@ -0,0 +1,1141 @@ + + + + + + 神话幻想世界 - Minecraft 克隆版 + + + +
+
+
正在生成神话世界...
+
+
+
神话幻想世界 | WASD 移动 | 鼠标环视
+
左键破坏 | 右键放置 | 1/2/3 选择方块
+
当前方块: 草地 | 位置: (0, 0, 0)
+
+
FPS: 60
+ + + + + + \ No newline at end of file diff --git a/outputs/mistral-medium-3-5/pelican-cycling.svg b/outputs/mistral-medium-3-5/pelican-cycling.svg new file mode 100644 index 0000000..8c75576 --- /dev/null +++ b/outputs/mistral-medium-3-5/pelican-cycling.svg @@ -0,0 +1,99 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + \ No newline at end of file