モデル比較

人気のAIモデルを性能・価格・特徴で比較

OpenAI

Arena Elo (テキスト): 1436

GPQA Diamond: 92.4%

SWE-Bench Verified: 80.0%

VS

GPQA Diamond

Claude Opus 4.7

Anthropic

GDPval-AA Elo: 1,753

SWE-bench Verified: 87.6%

SWE-bench Pro: 64.3%

GPT-5.2 詳細を見る Claude Opus 4.7 詳細を見る

OpenAI

Arena Elo (テキスト): 1436

GPQA Diamond: 92.4%

SWE-Bench Verified: 80.0%

VS

ARC-AGI-2

Google DeepMind

Arena Elo: 1486

GPQA Diamond: 91.9%

SWE-Bench Verified: 76.2%

GPT-5.2 詳細を見る Gemini 3.0 Pro 詳細を見る

Claude Opus 4.7

Anthropic

GDPval-AA Elo: 1,753

SWE-bench Verified: 87.6%

SWE-bench Pro: 64.3%

VS

Agentic

Google DeepMind

Arena Elo: 1486

GPQA Diamond: 91.9%

SWE-Bench Verified: 76.2%

Claude Opus 4.7 詳細を見る Gemini 3.0 Pro 詳細を見る

DeepSeek

Arena Elo: 1485

SWE-Bench Verified: 80.8%

入力価格: $0.28/1M tokens

VS

Price/Perf

OpenAI

Arena Elo (テキスト): 1436

GPQA Diamond: 92.4%

SWE-Bench Verified: 80.0%

DeepSeek V3.2 詳細を見る GPT-5.2 詳細を見る

Alibaba

SWE-bench Verified: 77.2%

AIME 2026: 94.1%

入力価格: $0.60 / 1M tokens

VS

Open-Weight

Llama-3-Namazu-405B

Sakana AI

MMLU (5-shot): 88.6%

HumanEval（コーディング）: 89.0%

入力価格: $2.40/M tokens

Qwen3.6-27B 詳細を見る Llama-3-Namazu-405B 詳細を見る

GPT-5.1 Codex Max

OpenAI

SWE-bench Verified (xhigh): 77.9%

SWE-Lancer IC SWE: 79.9%

Terminal-Bench 2.0: 58.1%

VS

Coding

Claude Opus 4.7

Anthropic

GDPval-AA Elo: 1,753

SWE-bench Verified: 87.6%

SWE-bench Pro: 64.3%

GPT-5.1 Codex Max 詳細を見る Claude Opus 4.7 詳細を見る

xAI

Chatbot Arena Elo: ~1493

IFBench (Instruction Following): 83%

Omniscience (Non-Hallucination): 78%

VS

Real-time

OpenAI

Arena Elo (テキスト): 1436

GPQA Diamond: 92.4%

SWE-Bench Verified: 80.0%

Grok 4.2 Beta 詳細を見る GPT-5.2 詳細を見る

Google DeepMind

Arena Elo: 1451

GPQA Diamond: 84.3%

LiveCodeBench v6: 80.0%

VS

Local Deploy

Alibaba

SWE-bench Verified: 77.2%

AIME 2026: 94.1%

入力価格: $0.60 / 1M tokens

Gemma 4 31B 詳細を見る Qwen3.6-27B 詳細を見る