AI models
AI Model Leaderboard — Benchmarks and Open-Source Trends
Gemini 4 Carbon (high) leads Artificial Analysis Intelligence Index v5.0 with 59 points (as of 2026-11-01). Claude Opus 5.5 follows at 58. On Hugging Face, idol tops the tracked trending list with 1,045 likes and 20,689 downloads.
Last updated · Benchmark snapshots have their own 'as of' dates shown per board.
第三方榜单快照，数字来自公开报道，未直接调用榜单平台 API；以来源链接为准。
Benchmark leaderboards
Artificial Analysis Intelligence Index v5.0
As of 2026-11-01 · Source: m.datalearner.com:8443/leaderboards/external/aa-quality-inde · m.163.com/dy/article/LARD0R3J05561FZL.html
| # | Model | Organization | Score | Note |
|---|---|---|---|---|
| 1 | Gemini 4 Carbon (high) | 59 | — | |
| 2 | Claude Opus 5.5 | Anthropic | 58 | — |
| 3 | GPT-6 Astra | OpenAI | 57 | — |
SuperCLUE 2026年11月通用大模型测评
As of 2026-11-05 · Source: www.chinaaet.com/article/3000181581 · m.163.com/dy/article/LAV1G0KT0511CPVM.html
| # | Model | Organization | Score | Note |
|---|---|---|---|---|
| 1 | Gemini-4-Argon | 73.62 | — | |
| 2 | GPT-6.1-Sol | OpenAI | 71.76 | — |
| 3 | Kimi-K3.1-Pro | 月之暗面 | 61.46 | 国内/开源第一 |
| 4 | DeepSeek-V4.1-Flash-1017 | DeepSeek | 58.95 | — |
| 5 | GLM-5.3 | 智谱 AI | 50.66 | — |
Arena.ai 盲测
As of 2026-11-03 · Source: www.szhzxw.cn/191768.html
| # | Model | Organization | Score | Note |
|---|---|---|---|---|
| 1 | Gemini 4 Carbon | — | Text Arena 第一；Agent Arena 第二 |
Hugging Face trending models
| # | Model | Trend score | Likes | Downloads | Pipeline |
|---|---|---|---|---|---|
| 1 | idol | 727 | 1,045 | 20,689 | image-to-3d |
| 2 | Kimi-K3.1-Pro | 519.5 | 1,602 | 22,571 | image-text-to-text |
| 3 | Qwen3.5-4B-3D-Coder | 420 | 843 | 221,506 | — |
| 4 | LTX-2.5 | 377 | 10,934 | 2,643,778 | image-to-video |
| 5 | Ripple-65B-Preview | 348 | 357 | 3,803 | — |
| 6 | numbat | 339 | 376 | 9,212 | — |
| 7 | Sulphur-3 | 337 | 402 | 997 | — |
| 8 | Qwen3.8-27B | 313 | 19,322 | 7,813,215 | image-text-to-text |
| 9 | NVIDIA-Nemotron-3.5-Transcribe-30B-A4B | 300.5 | 3,089 | 63,798 | automatic-speech-recognition |
| 10 | DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic | 297 | 351 | 1,369 | text-generation |
| 11 | Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NEO-CODER-MAX-MTP-GGUF | 284 | 2,632 | 1,545,313 | image-text-to-text |
| 12 | flux-3-image-klein-comfy | 253 | 2,663 | 1,143,397 | image-to-image |
| 13 | DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-GGUF | 244 | 362 | 117,016 | text-generation |
| 14 | DeepSeek-V4.1-Flash-1017-obliterated | 231 | 272 | 3,835 | text-generation |
| 15 | Pika-H3-Flex | 229 | 504 | 1,083 | image-to-video |
New on Hugging Face
| Model | Likes | Downloads | Created |
|---|---|---|---|
| Kimi-K3.1-Pro | 1,602 | 22,571 | Nov 2, 2026 |
| Ripple-65B-Preview | 357 | 3,803 | Nov 4, 2026 |
| Sulphur-3 | 402 | 997 | Oct 31, 2026 |
| DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic | 351 | 1,369 | Oct 30, 2026 |
| Pika-H3-Flex | 504 | 1,083 | Oct 29, 2026 |
AI news
- Google reveals new Gemini 4 Carbon AI model, narrowly beating Claude Opus 5.5 — frontier AI model is also the least likely to hallucinate, benchmarks claim Tom's Hardware · Nov 3, 2026
- Tencent Launches Hy4, Joining the Ranks of the World’s Leading Open-Source AI Models Tencent · Nov 4, 2026
- Ex-Googlers' new Apache 2.0 model Ripple-65B Preview claims the U.S. open coding lead on two major benchmarks VentureBeat · Nov 5, 2026
FAQ
Which AI model has the highest benchmark score?
Gemini 4 Carbon (high) ranks first on Artificial Analysis Intelligence Index v5.0 with 59 points, as of 2026-11-01. Scores are not directly comparable across different benchmarks because tasks and methods differ.
What is the Hugging Face trending list?
Hugging Face ranks models by a trending score that reflects recent likes, downloads and community activity. This page also shows lifetime likes and download counts, which move more slowly than the trending rank.
Are these benchmark numbers official?
They are snapshots of publicly published leaderboards, with the original sources linked above each table. The site does not re-run the benchmarks, so check the benchmark date before quoting a score.