| Modellslot | getestete Version | Zugriff | Local HW | AA Quality | Arena Elo | SWE Verified | Aider Poly | LiveBench | BFCL | RULER | MMMU | GPQA | Stand |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Claude Opus Anthropic | aus Quelle übernehmen | API | nein | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
Claude Sonnet Anthropic | aus Quelle übernehmen | API | nein | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
GPT Flagship OpenAI | aus Quelle übernehmen | API | nein | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
Gemini Pro Google | aus Quelle übernehmen | API | nein | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
Qwen Coder Alibaba / Open Weights | exakte Benchmark-Version | Weights | Server/GPU | missing | missing | missing | missing | missing | missing | missing | n/a | missing | Quelle |
DeepSeek Reasoner DeepSeek / Open Weights | exakte Benchmark-Version | API/W | Server/GPU | missing | missing | missing | missing | missing | missing | missing | n/a | missing | Quelle |
Llama Instruct Meta / Open Weights | exakte Benchmark-Version | Weights | Server/GPU | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
Mistral Large Mistral | exakte Benchmark-Version | API/W | Server/GPU | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
Gemma Google / Open Weights | exakte Benchmark-Version | Weights | Consumer+ | missing | missing | missing | missing | missing | missing | missing | missing | missing | Quelle |
Consumer Local z.B. MacBook / Ollama | konkretes GGUF/MLX | Local | MacBook | nicht vergleichb. | nicht vergleichb. | missing | missing | missing | missing | missing | n/a | missing | lokal |