Daten aktualisiert vor 3 MinutenQuellen:Code Arena·LiveCodeBench·Aider Polyglot
Live Benchmarks / Coding
Echte Code-Generierung, Repo-Level-Fixes und kompetitives Programmieren.
Aider Polyglot
Originalquelle ansehen →| Rank | Model | Score |
|---|---|---|
| 1 | gpt-5 (high)OpenAI | 88.0% correct |
| 2 | gpt-5 (medium)OpenAI | 86.7% correct |
| 3 | o3-pro (high)OpenAI | 84.9% correct |
| 4 | gemini-2.5-pro-preview-06-05 (32k think)Google DeepMind | 83.1% correct |
| 5 | gpt-5 (low)OpenAI | 81.3% correct |
| 6 | o3 (high)OpenAI | 81.3% correct |
| 7 | grok-4 (high)xAI | 79.6% correct |
| 8 | gemini-2.5-pro-preview-06-05 (default think)Google DeepMind | 79.1% correct |
| 9 | o3 (high) + gpt-4.1OpenAI | 78.2% correct |
| 10 | O3OpenAI | 76.9% correct |
| 11 | Gemini 2.5 Pro Preview 05-06Google DeepMind | 76.9% correct |
| 12 | DeepSeek-V3.2-Exp (Reasoner)DeepSeek | 74.2% correct |
| 13 | Gemini 2.5 Pro Preview 03-25Google DeepMind | 72.9% correct |
| 14 | claude-opus-4-20250514 (32k thinking)Anthropic | 72.0% correct |
| 15 | o4-mini (high)OpenAI | 72.0% correct |
| 16 | DeepSeek R1 (0528)DeepSeek | 71.4% correct |
| 17 | claude-opus-4-20250514 (no think)Anthropic | 70.7% correct |
| 18 | DeepSeek-V3.2-Exp (Chat)DeepSeek | 70.2% correct |
| 19 | claude-3-7-sonnet-20250219 (32k thinking tokens)Anthropic | 64.9% correct |
| 20 | DeepSeek R1 + claude-3-5-sonnet-20241022DeepSeek | 64.0% correct |
| 21 | o1-2024-12-17 (high)OpenAI | 61.7% correct |
| 22 | claude-sonnet-4-20250514 (32k thinking)Anthropic | 61.3% correct |
| 23 | claude-3-7-sonnet-20250219 (no thinking)Anthropic | 60.4% correct |
| 24 | o3-mini (high)OpenAI | 60.4% correct |
| 25 | Qwen3 235B A22B diff, no think, Alibaba APIAlibaba | 59.6% correct |
| 26 | Kimi K2Moonshot AI | 59.1% correct |
| 27 | DeepSeek R1DeepSeek | 56.9% correct |
| 28 | claude-sonnet-4-20250514 (no thinking)Anthropic | 56.4% correct |
| 29 | gemini-2.5-flash-preview-05-20 (24k think)Google DeepMind | 55.1% correct |
| 30 | DeepSeek V3 (0324)DeepSeek | 55.1% correct |
| 31 | Quasar AlphaUnknown | 54.7% correct |
| 32 | o3-mini (medium)OpenAI | 53.8% correct |
| 33 | Grok 3 BetaxAI | 53.3% correct |
| 34 | Optimus AlphaUnknown | 52.9% correct |
| 35 | Gpt 4.1OpenAI | 52.4% correct |
| 36 | Claude 3 5 Sonnet 20241022Anthropic | 51.6% correct |
| 37 | Grok 3 Mini Beta (high)xAI | 49.3% correct |
| 38 | DeepSeek Chat V3 (prev)DeepSeek | 48.4% correct |
| 39 | gemini-2.5-flash-preview-04-17 (default)Google DeepMind | 47.1% correct |
| 40 | chatgpt-4o-latest (2025-03-29)Unknown | 45.3% correct |
| 41 | Gpt 4.5 PreviewOpenAI | 44.9% correct |
| 42 | gemini-2.5-flash-preview-05-20 (no think)Google DeepMind | 44.0% correct |
| 43 | gpt-oss-120b (high)OpenAI | 41.8% correct |
| 44 | Qwen3 32BAlibaba | 40.0% correct |
| 45 | Gemini Exp 1206Google DeepMind | 38.2% correct |
| 46 | Gemini 2.0 Pro exp-02-05Google DeepMind | 35.6% correct |
| 47 | Grok 3 Mini Beta (low)xAI | 34.7% correct |
| 48 | O1 Mini 2024 09 12OpenAI | 32.9% correct |
| 49 | Gpt 4.1 MiniOpenAI | 32.4% correct |
| 50 | Claude 3 5 Haiku 20241022Anthropic | 28.0% correct |
| 51 | chatgpt-4o-latest (2025-02-15)Unknown | 27.1% correct |
| 52 | QwQ-32B + Qwen 2.5 Coder InstructUnknown | 26.2% correct |
| 53 | Gpt 4o 2024 08 06OpenAI | 23.1% correct |
| 54 | Gemini 2.0 Flash ExpGoogle DeepMind | 22.2% correct |
| 55 | Qwen Max 2025 01 25Alibaba | 21.8% correct |
| 56 | QwQ 32BUnknown | 20.9% correct |
| 57 | Gemini 2.0 Flash Thinking Exp 01 21Google DeepMind | 18.2% correct |
| 58 | Gpt 4o 2024 11 20OpenAI | 18.2% correct |
| 59 | DeepSeek Chat V2.5DeepSeek | 17.8% correct |
| 60 | Qwen2.5 Coder 32B InstructAlibaba | 16.4% correct |
| 61 | Llama 4 MaverickMeta | 15.6% correct |
| 62 | Yi LightningUnknown | 12.9% correct |
| 63 | Command A 03 2025 QualityUnknown | 12.0% correct |
| 64 | Codestral 25.01Mistral | 11.1% correct |
| 65 | Openhands Lm 32b v0.1Unknown | 10.2% correct |
| 66 | Gpt 4.1 NanoOpenAI | 8.9% correct |
| 67 | Qwen2.5 Coder 32B InstructAlibaba | 8.0% correct |
| 68 | Gemma 3 27b ItGoogle DeepMind | 4.9% correct |
| 69 | Gpt 4o Mini 2024 07 18OpenAI | 3.6% correct |
Code Arena
Originalquelle ansehen →| Rank | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 ThinkingAnthropic | 1548Elo |
| 2 | Claude Opus 4.6Anthropic | 1542Elo |
| 3 | Glm 5.1Z.ai | 1530Elo |
| 4 | Claude Sonnet 4.6Anthropic | 1521Elo |
| 5 | Claude Opus 4.5 ThinkingAnthropic | 1490Elo |
| 6 | Claude Opus 4.5Anthropic | 1466Elo |
| 7 | 1457Elo | |
| 8 | Gemini 3.1 ProGoogle | 1456Elo |
| 9 | Qwen 3.6 PlusAlibaba | 1453Elo |
| 10 | GLM-4.7Z.ai | 1439Elo |
| 11 | GLM-5Z.ai | 1439Elo |
| 12 | Gemini 3 ProGoogle | 1438Elo |
| 13 | 1437Elo | |
| 14 | Gemini 3 FlashGoogle | 1436Elo |
| 15 | MiMo V2 ProXiaomi | 1433Elo |
| 16 | Kimi K2.5 ThinkingMoonshot | 1429Elo |
| 17 | MiniMax M2.7MiniMax | 1425Elo |
| 18 | Kimi K2.5 InstantMoonshot | 1408Elo |
| 19 | 1407Elo | |
| 20 | GPT-5.2OpenAI | 1403Elo |
| 21 | 1393Elo | |
| 22 | GPT-5 MediumOpenAI | 1393Elo |
| 23 | MiniMax M2.5MiniMax | 1392Elo |
| 24 | MiniMax M2.1MiniMax | 1391Elo |
| 25 | GPT-5.1 MediumOpenAI | 1390Elo |
| 26 | 1390Elo | |
| 27 | Claude Sonnet 4.5 ThinkingAnthropic | 1388Elo |
| 28 | GPT-5.4 MiniOpenAI | 1388Elo |
| 29 | Qwen 3.5 397BAlibaba | 1386Elo |
| 30 | Claude Sonnet 4.5Anthropic | 1386Elo |
| 31 | Claude Opus 4.1Anthropic | 1385Elo |
| 32 | DeepSeek V3.2 ThinkingDeepSeek | 1368Elo |
| 33 | Qwen 3.5 122BAlibaba | 1365Elo |
| 34 | GLM-4.6Z.ai | 1354Elo |
| 35 | Qwen 3.5 27BAlibaba | 1344Elo |
| 36 | GPT-5.1OpenAI | 1339Elo |
| 37 | 1337Elo | |
| 38 | GPT-5.2 CodexOpenAI | 1335Elo |
| 39 | DeepSeek V3.2DeepSeek | 1330Elo |
| 40 | Kimi K2 TurboMoonshot | 1329Elo |
| 41 | Gpt 5.1 CodexOpenAI | 1329Elo |
| 42 | Claude Haiku 4.5Anthropic | 1315Elo |
| 43 | MiniMax M2MiniMax | 1304Elo |
| 44 | mimo-v2-flash (thinking)Xiaomi | 1300Elo |
| 45 | Deepseek v3.2 ExpDeepSeek | 1286Elo |
| 46 | Qwen 3 CoderAlibaba | 1281Elo |
| 47 | KAT Coder Pro v1Kwai | 1257Elo |
| 48 | Qwen3.5 35b A3bAlibaba | 1247Elo |
| 49 | Gpt 5.1 Codex MiniOpenAI | 1239Elo |
| 50 | Gemini 3.1 Flash LiteGoogle | 1237Elo |
| 51 | Qwen3.5 FlashAlibaba | 1236Elo |
| 52 | 1233Elo | |
| 53 | Mistral Large 3Mistral | 1222Elo |
| 54 | 1207Elo | |
| 55 | Gemini 2.5 ProGoogle | 1202Elo |
| 56 | Devstral 2Mistral | 1197Elo |
| 57 | Mercury 2Inception AI | 1166Elo |
| 58 | 1148Elo | |
| 59 | 1139Elo | |
| 60 | Devstral Medium 2507Mistral | 1091Elo |
LiveCodeBench
Originalquelle ansehen →| Rank | Model | Score |
|---|---|---|
| 1 | O4-Mini (High)OpenAI | 80.2pass@1 |
| 2 | O3 (High)OpenAI | 75.8pass@1 |
| 3 | O4-Mini (Medium)OpenAI | 74.2pass@1 |
| 4 | Gemini 2.5 Pro 06 05Google DeepMind | 73.6pass@1 |
| 5 | DeepSeek R1 0528DeepSeek | 73.1pass@1 |
| 6 | Gemini 2.5 Pro 05 06Google DeepMind | 71.8pass@1 |
| 7 | EXAONE 4.0 32BUnknown | 70.0pass@1 |
| 8 | OpenReasoning Nemotron 32BUnknown | 69.8pass@1 |
| 9 | O3-Mini-2025-01-31 (High)OpenAI | 67.4pass@1 |
| 10 | OpenCodeReasoning Nemotron 1.1 32BUnknown | 66.8pass@1 |
| 11 | Grok-3-Mini (High)xAI | 66.7pass@1 |
| 12 | O4-Mini (Low)OpenAI | 65.9pass@1 |
| 13 | Qwen3 235B A22BAlibaba | 65.9pass@1 |
| 14 | XBai O4 MediumUnknown | 65.0pass@1 |
| 15 | O3-Mini-2025-01-31 (Med)OpenAI | 63.0pass@1 |
| 16 | Gemini 2.5 Flash 05 20Google DeepMind | 61.9pass@1 |
| 17 | Gemini 2.5 Flash 04 17Google DeepMind | 60.6pass@1 |
| 18 | O3-Mini-2025-01-31 (Low)OpenAI | 57.0pass@1 |
| 19 | Claude-Opus-4 (Thinking)Anthropic | 56.6pass@1 |
| 20 | Claude-Sonnet-4 (Thinking)Anthropic | 55.9pass@1 |
| 21 | Claude Sonnet 4Anthropic | 47.1pass@1 |
| 22 | Claude Opus 4Anthropic | 46.9pass@1 |
| 23 | Claude 3.5 Sonnet 20241022Anthropic | 36.4pass@1 |
| 24 | GPT 4O 2024 08 06OpenAI | 29.5pass@1 |
| 25 | GPT 4 Turbo 2024 04 09OpenAI | 28.7pass@1 |
| 26 | GPT 4O Mini 2024 07 18OpenAI | 27.5pass@1 |
| 27 | DeepSeek v3DeepSeek | 27.2pass@1 |
| 28 | Claude 3 HaikuAnthropic | 20.2pass@1 |
WebDev Arena
Originalquelle ansehen →| Rank | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6Anthropic | 1467Elo |
| 2 | Claude Opus 4.6Anthropic | 1459Elo |
| 3 | GPT 5.4OpenAI | 1424Elo |
| 4 | Gemini 3.1 ProGoogle DeepMind | 1402Elo |
| 5 | Grok 4.20xAI | 1361Elo |
| 6 | DeepSeek V3.2DeepSeek | 1344Elo |
| 7 | Qwen 3.5Alibaba | 1326Elo |
| 8 | GLM-5Zhipu AI | 1311Elo |
| 9 | Kimi K2.5Moonshot AI | 1298Elo |
| 10 | MiniMax M2.5MiniMax | 1281Elo |
Verwandte Diskussion
In 1992 I built an online multiplayer game called Legends of Future Past. It ran on CompuServe, won an award from Computer Gaming World, and shut down on the last day of 1999. I was 19 when I made it. The source code didn't survive. What I
I want to be honest about something that happened to me because I think it is more common than people admit. Last month I hit a bug in a service I wrote myself two years ago. Network timeout issue, intermittent, only in prod. The kind of th
If it’s as good as they say it should be able to do it super easily. Have they just not thought about that?🤔
I got a 64gb memory mac about a month ago and I've been trying to find a model that is reasonably quick, decently good at coding, and doesn't overload my system. My test I've been running is having it create a doom style raycaster in html a
https://aisle.com/blog/ai-cybersecurity-after-mythos-the-jagged-frontier >We tested Anthropic Mythos's showcase vulnerabilities on small, cheap, open-weights models. They recovered much of the same analysis. AI cybersecurity capability i
I didn't realize but over a period of one year i have become overreliant on ChatGPT to write code, I am a second year PhD student and don't want to end up as someone with fake "coding skills" after I graduate. I hear people talk about it al
Benchmarks sind ein Ausgangspunkt, keine Antwort. Das richtige Modell hängt von Ihrem Workload, Budget und Ihren Integrations-Anforderungen ab – lassen Sie es uns gemeinsam herausfinden.