LLM Vergleich Tabelle
Juli 2024: Online Tabelle Vergleich
Leistungsfähigste aktuell zugängliche Modelle laut Chatbot Arena ⚔️ Benchmarks Leaderboard Anfang Juni 2024
Model | Arena Elo | Coding | Longer Query | German | Code-focused needle haystack | Organization | License | Knowledge Cutoff |
GPT-4o-2024-05-13 | 1287 | 1298 | 1304 | 1276 | 83 | OpenAI | Proprietary | 2023/10 |
Gemini-Advanced-0514 | 1267 | 1257 | 1260 | 1259 | Proprietary | Online | ||
Gemini-1.5-Pro-API-0514 | 1265 | 1272 | 1294 | 49 | Proprietary | 2023/11 | ||
Gemini-1.5-Pro-API-0409-Preview | 1257 | 1233 | 1246 | Proprietary | 2023/11 | |||
GPT-4-Turbo-2024-04-09 | 1256 | 1266 | 1266 | 1251 | 78 | OpenAI | Proprietary | 2023/12 |
GPT-4-1106-preview | 1251 | 1258 | 1248 | OpenAI | Proprietary | 2023/4 | ||
Claude 3 Opus | 1249 | 1252 | 1269 | 1249 | 60 | Anthropic | Proprietary | 2023/8 |
GPT-4-0125-preview | 1246 | 1246 | 1248 | OpenAI | Proprietary | 2023/12 | ||
Yi-Large-preview | 1239 | 1247 | 1246 | 01 AI | Proprietary | Unknown | ||
Gemini-1.5-Flash-API-0514 | 1231 | 1238 | 1223 | 36 | Proprietary | 2023/11 | ||
Yi-Large | 1222 | 1245 | 1223 | 1217 | 01 AI | Proprietary | Unknown | |
Bard (Gemini Pro) | 1208 | 1174 | 1131 | Proprietary | Online | |||
Llama-3-70b-Instruct | 1207 | 1202 | 1184 | 1160 | 35 | Meta | Llama 3 Community | 2023/12 |
Claude 3 Sonnet | 1201 | 1216 | 1221 | 1199 | Anthropic | Proprietary | 2023/8 | |
Reka-Core-20240501 | 1200 | 1192 | 1192 | 1192 | Reka AI | Proprietary | Unknown | |
Command R+ | 1189 | 1167 | 1213 | 1187 | 30 | Cohere | CC-BY-NC-4.0 | 2024/3 |
Qwen2-72B-Instruct | 1187 | 1189 | 1199 | 1136 | Alibaba | Qianwen LICENSE | 2024/6 | |
Mixtral-8x22b-Instruct-v0.1 | 1146 | 1153 | 1156 | 1130 | Mistral | Apache 2.0 | 2024/4 | |
GPT-3.5-Turbo-0613 | 1117 | 1137 | 1124 | 1117 | 36 | OpenAI | Proprietary | 2021/9 |
Siehe auch: Aider LLM Leaderboards (editing code); (NEUES FOSS Model): DeepSeek Coder V2 (no diff edit, hf, Aider 75%, HumanEval 90%); Artificial Analysis (Quality, Speed, Price, Context Size, etc.); Big Code Leaderboard;
Ältere Tabelle, freie/kleine Modelle ev. hinzugefügt, Source: Perplexity
GPT-4o | Gemini Advanced 1.5 Pro | Llama-3-70b-Instruct | MS Command R+ | Claude 3.0 Sonnet | GPT-3.5T | |
Zugang/Lizenz | Konto (limitiert) bzw. Abo $20/m | Abo 22€/m | meta Lic / Groq | CC-BY-NC-4.0 | Konto (No EU) | Konto |
Letzte Version/Vergleich | 2024-05-13 (2023-10) | 2024-05-14 (online) | 2023-12 | 2024-03-0x | 2023-07-12? | |
Areno Elo | 1287 | 1267 | 1208 | 1189 | 1202 | 1117 |
Arena Coding | 1299 | 1258 | 1202 | 1167 | 1216 | 1137 |
Kontext (Gesamt-Ein-Ausgabe) | 128k | 1000k | 8k | 128k | 100-200k | 4/16k |
Bilderkennung | Beschränkt | Beschränkt | ? | ? | x | x |
Coding | Ja | Ja | Ja | Ja | ||
in EU zugänglich | Ja | Ja | Ja | VPN | Ja | |
Needle in Haystack 1 2 | 35% | 41% | 36% | 42% | 35% | 35% |
Leistungsfähigste aktuell zugängliche Modelle laut Chatbot Arena ⚔️ Benchmarks Leaderboard Anfang März 2024, freie/kleine Modelle ev. hinzugefügt
GPT-4T | Gemini Pro | Mistral Medium | Mixtral 8x7b | Claude 3.0 Sonnet | GPT-3.5T | |
Zugang/Lizenz | Abo $20/m | Google Konto | Apache 2.0 | Konto (No EU) | Konto | |
Letzte Version/Vergleich | 2023-11-06 | 2024-01-30 | 2023-12-11 | 2023-12-11 | 2024-03-0x | 2023-07-12? |
Areno Elo | 1250 | 1200 | 1150 | 1115 | 1180 | 1115 |
Kontext (Gesamt-Ein-Ausgabe) | 4-128K | 2K | 1K | 0.5-32K | 200K | |
Bilderkennung | Beschränkt | Beschränkt | ? | ? | x | x |
Coding | Ja | Ja | Beschränkt | Beschränkt? | Ja | Ja |
in EU zugänglich | Ja | Ja | Ja | Ja | Ja | Ja |
Needle in Haystack 1 2 | 64 K 100% >100K <30% | Bad (Instruct)? | 32K >90% | 200K Opus >99% | ||
Claude 3.5 Sonnet: Erzeuge SVG mit Text aus PNG Karte
Wissenschaft/Science Research Help
Jan. 2024: In Englisch mit Hinweis zum Dienst und ob Login nötig ist. Von Andy Stapleton @ YT
- Consensus1: It is an AI-powered search engine that finds and summarizes scientific research papers. To use Consensus, you need to create an account1.
- Scholarly Assistant2: It is an AI assistant developed by Jenni AI to help with academic writing. It offers both free and paid features2.
- Paper Interpreter3: It is a tool that helps users understand academic papers. The usage details, including whether it’s free or requires a login, are not specified in the search results3.
- Scholar AI4: It is a website designed to help students with note-taking and learning. It offers both free and paid features4.
- Scholar GPT5: It is an AI model developed by OpenAI. It offers a subscription service for $20 per month, promising faster response times and priority access to new features. A free trial version is also available5.
- Academic Assistant Pro6: It acts as a professional academic assistant providing support with a professorial touch. The usage details, including whether it’s free or requires a login, are not specified in the search results6.
- Academic Research Reviewer7: It is a literature review generator that uses AI to provide comprehensive and meticulously curated literature reviews. The usage details, including whether it’s free or requires a login, are not specified in the search results7.
- Herisa or Urisa1: It seems there might be some confusion here. I couldn’t find an AI service named “Herisa”. However, URISA is an association for GIS professionals1. It doesn’t seem to be an AI tool for academic research.
- OpenRead234: OpenRead is an AI-driven platform that enables users to efficiently access and interact with papers. It offers a Paper Q&A tool to quickly answer any queries, a Paper Espresso feature to build and refine literature reviews, an AI-powered reading assistant, a low code editor, and an effective notes system3. It provides thousands of free and pre-made journal paper templates for easy kicking-off3. Most features on OpenRead can be unlocked for $5/month4.
- Explain Paper5678: ExplainPaper is a website that uses AI to summarize scientific research papers. Users can upload a paper, highlight confusing text, and get an explanation5. It makes understanding complex scientific papers easy8.
- PaperBrain91011: PaperBrain is a free AI tool that allows users to explore and understand research papers better10. It lets users search for a topic and it will show results relevant to their search10.
- Einblick121314: Einblick is an AI-native data notebook that writes and fixes code, plots beautiful charts, builds models, and much more12. It doesn’t specify whether it’s free or requires a login12.
- Tavaly15: is an AI tool for rapid insights and comprehensive research15.
LLM (Große Sprachmodelle – Textanfragen/Antworten)
- Juli 2023 Matt Wolfe Vergleich ChatGPT vs. Bard vs. Claude 2
Nicht inkludiert: Bing Chat, Perplexity AI