NEDEN ÖNEMLİ?
Analiz, hukuk, finans, teknik tasarım ve uzun doküman işlerinde modelin ilk cevabı değil, doğru sonuca sistematik şekilde ulaşması önemlidir.
NASIL OKUNMALI?
Muhakeme skoru yüksek modelleri kritik analizlerde tercih edin; fakat latency ve token tüketimi artabileceği için üretim bütçesiyle birlikte değerlendirin.
SINIRLAMA
Explicit reasoning modelleri çoğu zaman daha iyi sonuç verir, ancak daha yavaş ve pahalı olabilir. Kısa, basit görevlerde yüksek muhakeme modeli gereksiz maliyet yaratabilir.
Genel skora yaklaşık %17 etki MuSR LongBench v2 MRCRv2 ARC-AGI-2 BenchLM kaynağı BU KATEGORİDE ÖNE ÇIKANLAR
AJAN KOD GÖRSEL MUHAKEME BİLGİ TALİMAT ÇOK DİLLİ MATEMATİK BAĞLAM
LONG CONTEXT / MULTI-STEP / ABSTRACT
Muhakeme benchmark rehberi Modelin çok adımlı akıl yürütme, uzun bağlamda bilgi bulma, soyut problem çözme ve karmaşık karar zincirlerinde tutarlılık performansını ölçer.
AĞIRLIK / TİP
Genel skora yaklaşık %17 etki Neden önemli? Analiz, hukuk, finans, teknik tasarım ve uzun doküman işlerinde modelin ilk cevabı değil, doğru sonuca sistematik şekilde ulaşması önemlidir.
Nasıl okunmalı? Muhakeme skoru yüksek modelleri kritik analizlerde tercih edin; fakat latency ve token tüketimi artabileceği için üretim bütçesiyle birlikte değerlendirin.
Sınırlama Explicit reasoning modelleri çoğu zaman daha iyi sonuç verir, ancak daha yavaş ve pahalı olabilir. Kısa, basit görevlerde yüksek muhakeme modeli gereksiz maliyet yaratabilir.
# Model Sağlayıcı Muhakeme Ajan Kod Görsel Bağlam Fiyat I/O Kaynak 1 Qwen3.7 Plus Alibaba 84.8 87.7 76 1M $0.32 / $1.28 June 16, 2 + EKLE 2 Grok 4.1 xAI 76.6 81.7 97.8 1M API June 16, 2 + EKLE 3 GPT-5.4 Pro OpenAI 89.6 90.6 100 1.1M $30 / $180 June 16, 2 + EKLE 4 Gemini 3.1 Pro Google 81.3 93 84 1M $2 / $12 June 16, 2 + EKLE 5 GPT-5.5 OpenAI 95.9 82.6 58 1.1M $5 / $30 June 16, 2 + EKLE 6 GPT-5.4 OpenAI 84 87.2 59.6 1.1M $2.5 / $15 June 16, 2 + EKLE 7 Qwen3.7 Max Alibaba 86 91.6 - 1M $1.25 / $3.75 June 16, 2 + EKLE 8 GPT-5.3 Codex OpenAI 77.1 87.1 95.8 400K $1.75 / $14 June 16, 2 + EKLE 9 Claude Opus 4.7 (Adaptive) Anthropic 85.6 93.5 45 1M $5 / $25 June 16, 2 + EKLE 10 o1-preview OpenAI 87.1 78.8 67.5 200K $15 / $60 June 16, 2 + EKLE 11 Gemini 3 Pro Deep Think Google 93.3 73.4 100 2M API June 16, 2 + EKLE 12 GPT-5.1-Codex-Max OpenAI 75.5 75.2 90.1 400K $1.25 / $10 June 16, 2 + EKLE 13 Claude Opus 4.6 Anthropic 80.9 85.5 76.7 1M $5 / $25 June 16, 2 + EKLE 14 Grok 4.1 Fast xAI 61.1 59.8 89.7 1M $0.2 / $0.5 June 16, 2 + EKLE 15 GPT-5.2-Codex OpenAI 77.5 77.5 89 400K $1.75 / $14 June 16, 2 + EKLE 16 GLM-5 (Reasoning) Z.AI 77 69.1 72.6 200K $1 / $3.2 June 16, 2 + EKLE 17 GPT-5.2 OpenAI 59.2 78.7 81.5 400K $1.75 / $14 June 16, 2 + EKLE 18 Claude Sonnet 4.6 Anthropic 78.5 82 86.2 1M $3 / $15 June 16, 2 + EKLE 19 Gemini 3 Pro Google 70.4 72.4 81.2 2M $2 / $12 June 16, 2 + EKLE 20 Qwen3.5 397B (Reasoning) Alibaba 68.8 85.4 58.4 128K $0.6 / $3.6 June 16, 2 + EKLE 21 GPT-4.1 mini OpenAI 43 24.3 56.8 1.0M $0.4 / $1.6 June 16, 2 + EKLE 22 Gemini 3.5 Flash Google 94.5 77.3 80 1.0M $1.5 / $9 June 16, 2 + EKLE 23 GPT-4.1 OpenAI 55.2 54.6 63.9 1.0M $2 / $8 June 16, 2 + EKLE 24 GPT-5 (high) OpenAI 74.5 69.7 93.3 128K $1.25 / $10 June 16, 2 + EKLE 25 o1 OpenAI 56 48.2 58.9 200K $15 / $60 June 16, 2 + EKLE 26 GPT-5 (medium) OpenAI 67.7 77.7 90.6 400K $1.25 / $10 June 16, 2 + EKLE 27 GLM-4.7 Z.AI 53.4 72.5 58.2 203K $0.4 / $1.75 June 16, 2 + EKLE 28 o3-pro OpenAI 60.9 54.5 64.7 200K $20 / $80 June 16, 2 + EKLE 29 Qwen2.5-1M Alibaba 54.2 48.6 54.3 1M Ücretsiz June 16, 2 + EKLE 30 Claude Opus 4.5 Anthropic 73.2 75.6 61.2 200K $5 / $25 June 16, 2 + EKLE 31 Kimi K2.5 (Reasoning) Moonshot AI 61.9 85 67.8 128K $0.6 / $3 June 16, 2 + EKLE 32 o3-mini OpenAI 61.3 49.4 65.6 200K $1.1 / $4.4 June 16, 2 + EKLE 33 GPT-4.1 nano OpenAI 22.7 8.7 37.9 1.0M $0.1 / $0.4 June 16, 2 + EKLE 34 GPT-5.1 OpenAI 73.7 76 96.3 400K $1.25 / $10 June 16, 2 + EKLE 35 Gemini 3 Flash Google 44.3 47.5 72 1M $0.5 / $3 June 16, 2 + EKLE 36 Grok 4.20 xAI 54.2 76.4 47.6 2M $1.25 / $2.5 June 16, 2 + EKLE 37 Claude Sonnet 4.5 Anthropic 53.8 78.4 94.8 1M $3 / $15 June 16, 2 + EKLE 38 GLM-5.1 Z.AI 66 81.3 - 203K $0.98 / $3.08 June 16, 2 + EKLE 39 GLM-5 Z.AI 48.1 76.2 56 203K $0.6 / $1.92 June 16, 2 + EKLE 40 DeepSeekMath V2 DeepSeek 53 51.3 53.9 128K Ücretsiz June 16, 2 + EKLE 41 Qwen2.5-72B Alibaba 46.1 46.8 51.7 128K Ücretsiz June 16, 2 + EKLE 42 DeepSeek Coder 2.0 DeepSeek 59.3 60.9 36.2 128K API June 16, 2 + EKLE 43 Gemini 2.5 Pro Google 57.2 48.5 85.2 1.0M $1.25 / $10 June 16, 2 + EKLE 44 Gemini 3.1 Flash-Lite Google 41.1 25.8 58.3 1.0M $0.25 / $1.5 June 16, 2 + EKLE 45 Claude Haiku 4.5 Anthropic 44.1 53.2 73.6 200K $1 / $5 June 16, 2 + EKLE 46 Qwen3.5 397B Alibaba 53.4 67.5 65.1 128K $0.6 / $3.6 June 16, 2 + EKLE 47 DeepSeek LLM 2.0 DeepSeek 49.1 45.6 47.5 128K Ücretsiz June 16, 2 + EKLE 48 Nemotron 3 Super 100B NVIDIA 43.3 43 40 1M Ücretsiz June 16, 2 + EKLE 49 Claude 4.1 Opus Anthropic 38.6 56.4 77.3 200K $15 / $75 June 16, 2 + EKLE 50 MiMo-V2-Flash Xiaomi 56.9 68.6 68.7 256K Ücretsiz June 16, 2 + EKLE 51 Grok 4 xAI 52.8 76 73 128K API June 16, 2 + EKLE 52 o3 OpenAI 56.1 65.1 61.7 200K $2 / $8 June 16, 2 + EKLE 53 DeepSeek V3.2 (Thinking) DeepSeek 61.1 51.5 59 128K $0.55 / $2.19 June 16, 2 + EKLE 54 Claude 4 Sonnet Anthropic 43.6 53.4 75.5 200K $3 / $15 June 16, 2 + EKLE 55 Kimi K2.5 Moonshot AI 44.7 81 63.1 262K $0.375 / $2.025 June 16, 2 + EKLE 56 Gemini 1.5 Pro Google 33.6 14.9 65.5 2M $1.25 / $5 June 16, 2 + EKLE 57 Mistral Large 3 Mistral 37.2 37.5 67.7 128K $0.5 / $1.5 June 16, 2 + EKLE 58 Llama 3.1 405B Meta 31.2 36.5 43.8 128K Ücretsiz June 16, 2 + EKLE 59 Claude 3.5 Sonnet Anthropic 38.1 40.3 66.9 200K $3 / $15 June 16, 2 + EKLE 60 Claude 4.1 Opus Thinking Anthropic 24.7 51.3 65.7 200K API June 16, 2 + EKLE 61 o4-mini (high) OpenAI 38.5 45.5 50.3 200K $1.1 / $4.4 June 16, 2 + EKLE 62 DeepSeek V3.2 DeepSeek 49.8 57.3 50 131K $0.2288 / $0.3432 June 16, 2 + EKLE 63 GPT-4o OpenAI 38.6 24.4 62.2 128K $2.5 / $10 June 16, 2 + EKLE 64 Mistral Large 2 Mistral 32.3 32.6 41.1 128K API June 16, 2 + EKLE 65 Qwen3.6 Plus Alibaba 55.4 77.7 71.6 1M $0.325 / $1.95 June 16, 2 + EKLE 66 Nemotron 3 Ultra NVIDIA 50.5 80.4 - 1M $0.5 / $2.2 June 16, 2 + EKLE 67 Gemini 2.5 Flash Google 32.9 16.5 54 1.0M $0.3 / $2.5 June 16, 2 + EKLE 68 Grok Code Fast 1 xAI 39.9 61.6 21.2 256K $0.2 / $1.5 June 16, 2 + EKLE 69 Claude 3 Opus Anthropic 33.3 13.4 58.9 200K $15 / $75 June 16, 2 + EKLE 70 Llama 4 Scout Meta 17.1 - 36.4 10M $0.1 / $0.3 June 16, 2 + EKLE 71 Llama 3 70B Meta 21.9 2.6 25.8 128K Ücretsiz June 16, 2 + EKLE 72 Qwen3.5-27B Alibaba 50.2 70.3 65.1 262K $0.195 / $1.56 June 16, 2 + EKLE 73 Qwen3.5-122B-A10B Alibaba 55.2 70.7 65.2 262K $0.26 / $2.08 June 16, 2 + EKLE 74 Qwen3 235B 2507 (Reasoning) Alibaba 38.7 41.5 7.3 128K Ücretsiz June 16, 2 + EKLE 75 Qwen3.5-35B-A3B Alibaba 44.6 59.1 63.3 262K $0.14 / $1 June 16, 2 + EKLE 76 o1-pro OpenAI 16.1 14.5 19.3 200K $150 / $600 June 16, 2 + EKLE 77 GPT-OSS 120B OpenAI 34.3 21.4 18.9 131K $0.039 / $0.18 June 16, 2 + EKLE 78 Grok 3 [Beta] xAI 21 28.2 9.4 128K API June 16, 2 + EKLE 79 Nemotron 3 Nano 30B NVIDIA 19.7 16.8 12.5 32K Ücretsiz June 16, 2 + EKLE 80 DeepSeek V3.1 (Reasoning) DeepSeek 32.8 21 6.2 128K Ücretsiz June 16, 2 + EKLE 81 Qwen3 235B 2507 Alibaba 16.3 22.8 6.5 128K Ücretsiz June 16, 2 + EKLE 82 DeepSeek-R1 DeepSeek 31.1 26.4 17 128K $0.55 / $2.19 June 16, 2 + EKLE 83 GPT-4o mini OpenAI 38.5 90.9 41.2 128K $0.15 / $0.6 June 16, 2 + EKLE 84 GLM-4.5 Z.AI 22.9 7.8 5.2 131K $0.6 / $2.2 June 16, 2 + EKLE 85 DeepSeek V3 DeepSeek - 28.5 - 131K $0.2002 / $0.8001 June 16, 2 + EKLE 86 DeepSeek V3.1 DeepSeek 25.9 15.9 3.7 164K $0.21 / $0.79 June 16, 2 + EKLE 87 Claude Mythos 5 Anthropic - 100 100 98.9 1M $10 / $50 June 16, 2 + EKLE 88 Claude Fable 5 Anthropic - 100 100 79 1M $10 / $50 June 16, 2 + EKLE 89 GLM-5.2 Z.AI - 100 85.6 - 1.0M $1.4 / $4.4 June 16, 2 + EKLE 90 Claude Opus 4.8 Anthropic - 96 98 66.4 1M $5 / $25 June 16, 2 + EKLE 91 DeepSeek V4 Pro (Max) DeepSeek - 87.1 89.7 - 1M $1.74 / $3.48 June 16, 2 + EKLE 92 DeepSeek V4 Pro (High) DeepSeek - 81.5 86 - 1.0M $0.435 / $0.87 June 16, 2 + EKLE 93 Kimi K2.6 Moonshot AI - 82 89.8 71.3 262K $0.68 / $3.41 June 16, 2 + EKLE 94 MiniMax M3 MiniMax - 85 84.6 48.1 1.0M $0.3 / $1.2 June 16, 2 + EKLE 95 DeepSeek V4 Flash (Max) DeepSeek - 68.4 82.2 - 1M $0.14 / $0.28 June 16, 2 + EKLE 96 Qwen3.6-27B Alibaba - 68.7 77.9 65 262K $0.2885 / $3.17 June 16, 2 + EKLE 97 DeepSeek V4 Flash (High) DeepSeek - 60.1 82.5 - 1M $0.14 / $0.28 June 16, 2 + EKLE 98 DeepSeek V4 Pro DeepSeek - 65.6 68.4 - 1M $1.74 / $3.48 June 16, 2 + EKLE 99 MAI-Thinking-1 Microsoft - 31.2 84.5 - 256K API June 16, 2 + EKLE 100 Qwen3.6-35B-A3B Alibaba - 53.1 72.2 64 262K $0.15 / $1 June 16, 2 + EKLE 101 DeepSeek V4 Flash DeepSeek - 47.2 62 - 1.0M $0.09 / $0.18 June 16, 2 + EKLE 102 MiniMax M2.7 MiniMax - 47.8 54 - 205K $0.25 / $1 June 16, 2 + EKLE 103 Kimi K2 Moonshot AI - 30 53.4 - 128K $0.6 / $2.5 June 16, 2 + EKLE 104 Sarvam 105B Sarvam - 28.9 36.9 - 128K Ücretsiz June 16, 2 + EKLE 105 MiniCPM5-1B OpenBMB - - - - 131K API June 16, 2 + EKLE 106 DBRX Instruct Databricks - 8.3 60 - 32K Ücretsiz June 16, 2 + EKLE 107 Phi-4 Microsoft - 11.5 48.1 18.9 16K $0.065 / $0.14 June 16, 2 + EKLE
Veriler Supabase kataloğundaki BenchLM.ai ve OpenRouter kayıtlarından okunur. Fiyat ve skorlar sağlayıcı güncellemeleriyle değişebilir. Kaynaklar: BenchLM.ai · OpenRouter