News-Tracker

Modeller

AI & LLM · daglig briefing

Ændringer siden forrige snapshot

Ny i indekset

  • Gemini 3.7 Flash (medium) er ny i Artificial Analysis' indeks (intelligence 39.6, Google).
  • Gemini 3.5 Flash (medium) er ny i Artificial Analysis' indeks (intelligence 33.6, Google).
  • DeepSeek V4 Pro (Reasoning, Max Effort) er ny i Artificial Analysis' indeks (intelligence 30.9, DeepSeek).
  • GPT-5.5 Instant (June 2026) er ny i Artificial Analysis' indeks (intelligence 26.8, OpenAI).
  • DeepSeek V4 Flash (Reasoning, High Effort) er ny i Artificial Analysis' indeks (intelligence 24.8, DeepSeek).
  • Qwen3.5 397B A17B (Non-reasoning) er ny i Artificial Analysis' indeks (intelligence 21.4, Alibaba).
  • Qwen3.5 122B A10B (Non-reasoning) er ny i Artificial Analysis' indeks (intelligence 17.7, Alibaba).
  • gpt-oss-20b (low) er ny i Artificial Analysis' indeks (intelligence 10, OpenAI).
  • Qwen3 32B (Non-reasoning) er ny i Artificial Analysis' indeks (intelligence 7.3, Alibaba).
  • Qwen3 14B (Non-reasoning) er ny i Artificial Analysis' indeks (intelligence 6.7, Alibaba).
  • DeepSeek-V2.5 er ny i Artificial Analysis' indeks (intelligence 6.6, DeepSeek).
  • Qwen3.5 0.8B (Reasoning) er ny i Artificial Analysis' indeks (intelligence 6.1, Alibaba).
  • Qwen3 8B (Non-reasoning) er ny i Artificial Analysis' indeks (intelligence 6, Alibaba).
  • Mistral Small (Sep '24) er ny i Artificial Analysis' indeks (intelligence 5.8, Mistral).
  • Exaone 4.0 1.2B (Reasoning) er ny i Artificial Analysis' indeks (intelligence 5.3, LG AI Research).
  • Qwen3 1.7B (Reasoning) er ny i Artificial Analysis' indeks (intelligence 5.2, Alibaba).

Score ændret

  • GPT-5.4 nano (xhigh): intelligence 30.9 → 21.2 (-9.7).
  • DeepSeek V3.1 Terminus (Reasoning): intelligence 23.5 → 15.4 (-8.1).
  • Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort): intelligence 38.5 → 30.5 (-8.0).
  • Mistral Medium 3.5: intelligence 22.7 → 14.9 (-7.8).
  • Qwen3.8 27B (xhigh): intelligence 41.4 → 33.9 (-7.5).
  • GPT-5.4 mini (xhigh): intelligence 31.9 → 24.6 (-7.3).
  • Quasar 438B (max, based on GLM-5.2): intelligence 34.2 → 27.1 (-7.1).
  • Qwen3.6 27B (Reasoning): intelligence 29 → 21.9 (-7.1).
  • Muse Spark 1.2 (xhigh): intelligence 46.8 → 39.8 (-7.0).
  • GPT-5.5 (xhigh): intelligence 45.6 → 38.6 (-7.0).
  • Muse Spark 1.1 (xhigh): intelligence 41.2 → 34.3 (-6.9).
  • MiniMax-M2.7: intelligence 30.1 → 23.2 (-6.9).
  • Gemma 4 31B (Reasoning): intelligence 22.2 → 15.4 (-6.8).
  • Qwen3.8 2.4T A95B: intelligence 46.7 → 40 (-6.7).
  • Claude Sonnet 5 (Adaptive Reasoning, Max Effort): intelligence 45.1 → 38.4 (-6.7).
  • Qwen3.7 Max: intelligence 36.6 → 29.9 (-6.7).
  • Inkling (xhigh): intelligence 32.2 → 25.5 (-6.7).
  • Qwen3.8 Max: intelligence 46.9 → 40.3 (-6.6).
  • Hy3: intelligence 32.4 → 25.8 (-6.6).
  • Kimi K3 (max): intelligence 50.2 → 43.8 (-6.4).
  • Grok 4.5 (high): intelligence 45.5 → 39.1 (-6.4).
  • Kimi K2.7 Code: intelligence 32.7 → 26.3 (-6.4).
  • Ring-2.6-1T: intelligence 23.7 → 17.3 (-6.4).
  • DeepSeek V4 Flash 0731 (Reasoning, Max Effort): intelligence 40.8 → 34.5 (-6.3).
  • Gemini 3.1 Pro Preview: intelligence 36.7 → 30.4 (-6.3).
  • Muse Glimmer (high): intelligence 24.4 → 18.1 (-6.3).
  • Grok 4.6 (high): intelligence 50.6 → 44.4 (-6.2).
  • MiMo-V2.5-Pro: intelligence 32.6 → 26.4 (-6.2).
  • MiniMax-M3: intelligence 35.7 → 29.6 (-6.1).
  • Inkling Small: intelligence 32.2 → 26.1 (-6.1).
  • Qwen3.7 Plus: intelligence 31.9 → 25.8 (-6.1).
  • Claude 4.5 Sonnet (Reasoning): intelligence 27.3 → 21.2 (-6.1).
  • LongCat 2.0: intelligence 25.8 → 19.7 (-6.1).
  • Gemini 3.6 Flash (high): intelligence 40.3 → 34.3 (-6.0).
  • Gemini 3.8 Flash (high): intelligence 47.1 → 41.2 (-5.9).
  • GPT-5.6 Luna (max): intelligence 43.4 → 37.5 (-5.9).
  • Nemotron 3 Ultra 550B A55B (Reasoning): intelligence 29.3 → 23.4 (-5.9).
  • MiMo-V2.5: intelligence 28.2 → 22.3 (-5.9).
  • Claude Opus 4.8 (Adaptive Reasoning, Max Effort): intelligence 47.8 → 42 (-5.8).
  • DeepSeek V4 Pro 0813 (Reasoning, Max Effort): intelligence 42.1 → 36.3 (-5.8).
  • DeepSeek V4 Flash Vision (Reasoning, Max Effort): intelligence 40.7 → 35 (-5.7).
  • Granite 4.1 8B: intelligence 1.1 → 6.6 (+5.5).
  • Sarvam 30B (high): intelligence 1.1 → 6.6 (+5.5).
  • DeepSeek R1 Distill Llama 8B: intelligence 1 → 6.5 (+5.5).
  • Mistral Saba: intelligence 1 → 6.5 (+5.5).
  • Olmo 3.1 32B Instruct: intelligence 1 → 6.5 (+5.5).
  • Olmo 3 32B Think: intelligence 1 → 6.5 (+5.5).
  • Llama 3.1 Instruct 70B: intelligence 1.2 → 6.6 (+5.4).
  • Llama 3.2 Instruct 90B (Vision): intelligence 1 → 6.4 (+5.4).
  • Qwen2.5 Turbo: intelligence 1 → 6.4 (+5.4).
  • R1 1776: intelligence 1 → 6.4 (+5.4).
  • Reka Flash (Sep '24): intelligence 1 → 6.4 (+5.4).
  • Solar Mini: intelligence 1 → 6.4 (+5.4).
  • GPT-4o mini: intelligence 1.4 → 6.7 (+5.3).
  • Mistral Small 3: intelligence 1.4 → 6.7 (+5.3).
  • Nova Lite: intelligence 1.4 → 6.7 (+5.3).
  • Grok-1: intelligence 1 → 6.3 (+5.3).
  • Phi-4 Mini Instruct: intelligence 1 → 6.3 (+5.3).
  • Qwen2 Instruct 72B: intelligence 1 → 6.3 (+5.3).
  • GPT-4: intelligence 1.5 → 6.7 (+5.2).
  • Gemini 1.5 Flash-8B: intelligence 1 → 6.2 (+5.2).
  • Qwen2.5 Coder Instruct 32B: intelligence 1.6 → 6.7 (+5.1).
  • Qwen3 4B 2507 Instruct: intelligence 1.6 → 6.7 (+5.1).
  • DeepHermes 3 - Mistral 24B Preview (Non-reasoning): intelligence 1 → 6.1 (+5.1).
  • Jamba 1.7 Large: intelligence 1 → 6.1 (+5.1).
  • Nemotron 3 Super 120B A12B (Reasoning): intelligence 18.6 → 13.6 (-5.0).
  • Grok Beta: intelligence 1.9 → 6.9 (+5.0).
  • Qwen2.5 Instruct 32B: intelligence 1.9 → 6.9 (+5.0).
  • DeepSeek-Coder-V2: intelligence 1 → 6 (+5.0).
  • Granite 4.0 H Small: intelligence 1 → 6 (+5.0).
  • Hermes 3 - Llama-3.1 70B: intelligence 1 → 6 (+5.0).
  • Jamba 1.5 Large: intelligence 1 → 6 (+5.0).
  • Jamba 1.6 Large: intelligence 1 → 6 (+5.0).
  • OLMo 2 32B: intelligence 1 → 6 (+5.0).
  • Qwen3 Omni 30B A3B Instruct: intelligence 1 → 6 (+5.0).
  • Gemini 3.5 Flash-Lite: intelligence 27.6 → 22.7 (-4.9).
  • Claude 4.5 Haiku (Reasoning): intelligence 22.5 → 17.6 (-4.9).
  • Nova Pro: intelligence 2.1 → 7 (+4.9).
  • Llama 3.1 Instruct 8B: intelligence 2 → 6.9 (+4.9).
  • Claude 3 Sonnet: intelligence 1 → 5.9 (+4.9).
  • Granite 4.1 3B: intelligence 1 → 5.9 (+4.9).
  • LFM2 24B A2B: intelligence 1 → 5.9 (+4.9).
  • Nova Micro: intelligence 1 → 5.9 (+4.9).
  • Phi-4: intelligence 1 → 5.9 (+4.9).
  • Muse Spark 1.3 (max): intelligence 53 → 48.2 (-4.8).
  • Llama 3.1 Nemotron Instruct 70B: intelligence 2.1 → 6.9 (+4.8).
  • Qwen3.5 2B (Reasoning): intelligence 2.1 → 6.9 (+4.8).
  • Gemini 1.0 Ultra: intelligence 1 → 5.8 (+4.8).
  • Gemma 3n E4B Instruct Preview (May '25): intelligence 1 → 5.8 (+4.8).
  • Mistral Large (Feb '24): intelligence 1 → 5.8 (+4.8).
  • Phi-3 Mini Instruct 3.8B: intelligence 1 → 5.8 (+4.8).
  • Phi-4 Multimodal Instruct: intelligence 1 → 5.8 (+4.8).
  • Qwen2.5 Coder Instruct 7B : intelligence 1 → 5.8 (+4.8).
  • Gemini 1.5 Flash (Sep '24): intelligence 2.4 → 7.1 (+4.7).
  • Grok 2 (Dec '24): intelligence 2.4 → 7.1 (+4.7).
  • GPT-4 Turbo: intelligence 2.3 → 7 (+4.7).
  • Qwen3 VL 4B (Reasoning): intelligence 2.3 → 7 (+4.7).
  • Jamba Reasoning 3B: intelligence 1 → 5.7 (+4.7).
  • Llama 2 Chat 7B: intelligence 1 → 5.7 (+4.7).
  • Llama 3.2 Instruct 3B: intelligence 1 → 5.7 (+4.7).
  • MiniCPM-V 4.6 1.3B: intelligence 1 → 5.7 (+4.7).
  • Mixtral 8x22B Instruct: intelligence 1 → 5.7 (+4.7).
  • Qwen1.5 Chat 110B: intelligence 1 → 5.7 (+4.7).
  • Qwen3 VL 4B Instruct: intelligence 1 → 5.7 (+4.7).
  • GPT-5.2 (xhigh): intelligence 35 → 30.4 (-4.6).
  • Ring-flash-2.0: intelligence 2.6 → 7.2 (+4.6).
  • Olmo 3.1 32B Think: intelligence 2.5 → 7.1 (+4.6).
  • Pixtral Large: intelligence 2.5 → 7.1 (+4.6).
  • Claude 2.1: intelligence 1 → 5.6 (+4.6).
  • Claude 3 Haiku: intelligence 1 → 5.6 (+4.6).
  • Molmo 7B-D: intelligence 1 → 5.6 (+4.6).
  • OLMo 2 7B: intelligence 1 → 5.6 (+4.6).
  • Olmo 3 7B Think: intelligence 1 → 5.6 (+4.6).
  • Reka Flash 3: intelligence 1 → 5.6 (+4.6).
  • GPT-5.6 Terra (max): intelligence 46.8 → 42.3 (-4.5).
  • Claude Opus 4.6 (Adaptive Reasoning, Max Effort): intelligence 36.4 → 31.9 (-4.5).
  • Kimi K2.6: intelligence 35.8 → 31.3 (-4.5).
  • Muse Spark: intelligence 35.8 → 31.3 (-4.5).
  • Apodex 1.1: intelligence 34.9 → 30.4 (-4.5).
  • Claude Opus 4.5 (Reasoning): intelligence 33.6 → 29.1 (-4.5).
  • MiMo-V2-Pro: intelligence 33.1 → 28.6 (-4.5).
  • GPT-5.2 Codex (xhigh): intelligence 33 → 28.5 (-4.5).
  • Qwen3.6 Max Preview: intelligence 32.9 → 28.4 (-4.5).
  • Nex-N2-Pro: intelligence 32.7 → 28.2 (-4.5).
  • Solar Pro 4: intelligence 32.7 → 28.2 (-4.5).
  • GLM-5 (Reasoning): intelligence 32.4 → 27.9 (-4.5).
  • GLM-5.1 (Reasoning): intelligence 31.9 → 27.4 (-4.5).
  • JT-4.1 Flash 236B A21B: intelligence 31.8 → 27.3 (-4.5).
  • Grok Build 0.1 0616: intelligence 31.7 → 27.2 (-4.5).
  • Qwen3.6 Plus: intelligence 31.5 → 27 (-4.5).
  • Grok 4.3 (high): intelligence 29.9 → 25.4 (-4.5).
  • Qwen3 VL 8B Instruct: intelligence 2.8 → 7.3 (+4.5).
  • LFM2.5-8B-A1B: intelligence 2.7 → 7.2 (+4.5).
  • Qwen3 4B (Reasoning): intelligence 2.7 → 7.2 (+4.5).
  • Llama 3.1 Tulu3 405B: intelligence 2.7 → 7.2 (+4.5).
  • Claude 2.0: intelligence 1 → 5.5 (+4.5).
  • DeepSeek R1 Distill Qwen 1.5B: intelligence 1 → 5.5 (+4.5).
  • DeepSeek-V2-Chat: intelligence 1 → 5.5 (+4.5).
  • GPT-3.5 Turbo: intelligence 1 → 5.5 (+4.5).
  • Ling-mini-2.0: intelligence 1 → 5.5 (+4.5).
  • Llama 3 Instruct 70B: intelligence 1 → 5.5 (+4.5).
  • Mistral Medium: intelligence 1 → 5.5 (+4.5).
  • Motif 3: intelligence 38 → 33.6 (-4.4).
  • GPT-5.3 Codex (xhigh): intelligence 36.9 → 32.5 (-4.4).
  • Gemini 3 Pro Preview (high): intelligence 32.4 → 28 (-4.4).
  • Agnes 2.5 Pro Alpha: intelligence 31.2 → 26.8 (-4.4).
  • GLM-5-Turbo: intelligence 31 → 26.6 (-4.4).
  • Gemini 3 Flash Preview (Reasoning): intelligence 30.7 → 26.3 (-4.4).
  • GPT-5.1 (high): intelligence 29.1 → 24.7 (-4.4).
  • Kimi Linear 48B A3B Instruct: intelligence 2.9 → 7.3 (+4.4).
  • Llama 3.1 Instruct 405B: intelligence 2.9 → 7.3 (+4.4).
  • Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning): intelligence 2.9 → 7.3 (+4.4).
  • Arctic Instruct: intelligence 1 → 5.4 (+4.4).
  • LFM 40B: intelligence 1 → 5.4 (+4.4).
  • Llama 3.2 Instruct 11B (Vision): intelligence 1 → 5.4 (+4.4).
  • PALM-2: intelligence 1 → 5.4 (+4.4).
  • Qwen Chat 72B: intelligence 1 → 5.4 (+4.4).
  • GLM-5.3-Flash: intelligence 46.2 → 41.9 (-4.3).
  • Agnes 2.5 Pro Beta: intelligence 39.5 → 35.2 (-4.3).
  • Grok 4.20 0309 v2 (Reasoning): intelligence 30 → 25.7 (-4.3).
  • Grok 4.20 0309 (Reasoning): intelligence 29.5 → 25.2 (-4.3).
  • MiMo-V2-Omni-0327: intelligence 29.4 → 25.1 (-4.3).
  • GPT-5 Codex (high): intelligence 29.2 → 24.9 (-4.3).
  • Solar Open2 250B: intelligence 29 → 24.7 (-4.3).
  • NVIDIA Nemotron 3 Nano 4B: intelligence 3.1 → 7.4 (+4.3).
  • Command-R+ (Apr '24): intelligence 1 → 5.3 (+4.3).
  • DBRX Instruct: intelligence 1 → 5.3 (+4.3).
  • DeepSeek Coder V2 Lite Instruct: intelligence 1 → 5.3 (+4.3).
  • DeepSeek LLM 67B Chat (V1): intelligence 1 → 5.3 (+4.3).
  • Gemini 1.0 Pro: intelligence 1 → 5.3 (+4.3).
  • Llama 2 Chat 13B: intelligence 1 → 5.3 (+4.3).
  • Llama 2 Chat 70B: intelligence 1 → 5.3 (+4.3).
  • OpenChat 3.5 (1210): intelligence 1 → 5.3 (+4.3).
  • Sarvam M (Reasoning): intelligence 1 → 5.3 (+4.3).
  • GPT-5.6 Sol (max): intelligence 51.3 → 47.1 (-4.2).
  • MiMo-V2-Omni: intelligence 28.1 → 23.9 (-4.2).
  • GPT-5.1 Codex (high): intelligence 27.9 → 23.7 (-4.2).
  • NVIDIA Nemotron Nano 12B v2 VL (Reasoning): intelligence 3.3 → 7.5 (+4.2).
  • Gemini 2.0 Flash-Lite (Feb '25): intelligence 3.2 → 7.4 (+4.2).
  • Granite 4.1 30B: intelligence 3.2 → 7.4 (+4.2).
  • NVIDIA Nemotron Nano 9B V2 (Reasoning): intelligence 3.2 → 7.4 (+4.2).
  • Granite 4.0 H 1B: intelligence 1 → 5.2 (+4.2).
  • Jamba 1.5 Mini: intelligence 1 → 5.2 (+4.2).
  • Jamba 1.6 Mini: intelligence 1 → 5.2 (+4.2).
  • Jamba 1.7 Mini: intelligence 1 → 5.2 (+4.2).
  • LFM2 2.6B: intelligence 1 → 5.2 (+4.2).
  • LFM2.5-1.2B-Instruct: intelligence 1 → 5.2 (+4.2).
  • LFM2.5-1.2B-Thinking: intelligence 1 → 5.2 (+4.2).
  • Olmo 3 7B Instruct: intelligence 1 → 5.2 (+4.2).
  • GLM 5V Turbo (Reasoning): intelligence 27.6 → 23.5 (-4.1).
  • Kimi K2.5 (Reasoning): intelligence 27.6 → 23.5 (-4.1).
  • GPT-5 (high): intelligence 27.1 → 23 (-4.1).
  • Claude 4.1 Opus (Reasoning): intelligence 26.9 → 22.8 (-4.1).
  • A.X-K2: intelligence 26.8 → 22.7 (-4.1).
  • Hy3-preview (Reasoning): intelligence 26.8 → 22.7 (-4.1).
  • GLM-4.5V (Reasoning): intelligence 3.5 → 7.6 (+4.1).
  • Mistral Large 2 (Nov '24): intelligence 3.5 → 7.6 (+4.1).
  • ERNIE 4.5 300B A47B: intelligence 3.4 → 7.5 (+4.1).
  • Hermes 4 - Llama-3.1 405B (Reasoning): intelligence 3.4 → 7.5 (+4.1).
  • Llama 3.1 Nemotron Ultra 253B v1 (Reasoning): intelligence 3.4 → 7.5 (+4.1).
  • Qwen3 30B A3B 2507 Instruct: intelligence 3.4 → 7.5 (+4.1).
  • Apertus 70B Instruct: intelligence 1 → 5.1 (+4.1).
  • DeepHermes 3 - Llama-3.1 8B Preview (Non-reasoning): intelligence 1 → 5.1 (+4.1).
  • Gemma 3 270M: intelligence 1 → 5.1 (+4.1).
  • Granite 4.0 Micro: intelligence 1 → 5.1 (+4.1).
  • Mixtral 8x7B Instruct: intelligence 1 → 5.1 (+4.1).
  • Qwen3.5 27B (Reasoning): intelligence 26.9 → 22.9 (-4.0).
  • MiniMax-M2.5: intelligence 26.8 → 22.8 (-4.0).
  • Grok 4: intelligence 26.5 → 22.5 (-4.0).
  • MiMo-V2-Flash (Feb 2026): intelligence 26.4 → 22.4 (-4.0).
  • GLM-4.7 (Reasoning): intelligence 26.2 → 22.2 (-4.0).
  • Qwen3 Coder Next: intelligence 14.1 → 10.1 (-4.0).
  • Llama 3.3 Instruct 70B: intelligence 3.7 → 7.7 (+4.0).
  • QwQ 32B-Preview: intelligence 3.6 → 7.6 (+4.0).
  • Qwen3 30B A3B (Reasoning): intelligence 3.6 → 7.6 (+4.0).
  • Sonar Pro: intelligence 3.6 → 7.6 (+4.0).
  • Claude Instant: intelligence 1 → 5 (+4.0).
  • Granite 4.0 1B: intelligence 1 → 5 (+4.0).
  • Llama 65B: intelligence 1 → 5 (+4.0).
  • Mistral 7B Instruct: intelligence 1 → 5 (+4.0).
  • Molmo2-8B: intelligence 1 → 5 (+4.0).
  • Qwen Chat 14B: intelligence 1 → 5 (+4.0).
  • K2 Horizon 375B A23B: intelligence 37.8 → 33.9 (-3.9).
  • Qwen3.6 35B A3B (Reasoning): intelligence 26.2 → 22.3 (-3.9).
  • Kimi K2 Thinking: intelligence 25.9 → 22 (-3.9).
  • G9v3-39A5B: intelligence 25.7 → 21.8 (-3.9).
  • KAT Coder Pro V2: intelligence 25.6 → 21.7 (-3.9).
  • Sonar: intelligence 3.8 → 7.7 (+3.9).
  • Step3 VL 10B: intelligence 3.8 → 7.7 (+3.9).
  • Granite 3.3 8B (Non-reasoning): intelligence 1 → 4.9 (+3.9).
  • LFM2 8B A1B: intelligence 1 → 4.9 (+3.9).
  • GPT-5.4 (xhigh): intelligence 42.8 → 39 (-3.8).
  • o3-pro: intelligence 25.7 → 21.9 (-3.8).
  • DeepSeek V3.2 (Reasoning): intelligence 25.3 → 21.5 (-3.8).
  • Ling 3.0 Tiny: intelligence 15.7 → 11.9 (-3.8).
  • Gemma 4 E2B (Reasoning): intelligence 4 → 7.8 (+3.8).
  • Qwen3 Omni 30B A3B (Reasoning): intelligence 4 → 7.8 (+3.8).
  • Qwen2.5 Instruct 72B: intelligence 3.9 → 7.7 (+3.8).
  • Apertus 8B Instruct: intelligence 1 → 4.8 (+3.8).
  • Gemma 3 1B Instruct: intelligence 1 → 4.8 (+3.8).
  • Gemma 3 4B Instruct: intelligence 1 → 4.8 (+3.8).
  • Gemma 3n E2B Instruct: intelligence 1 → 4.8 (+3.8).
  • Gemma 3n E4B Instruct: intelligence 1 → 4.8 (+3.8).
  • Granite 4.0 350M: intelligence 1 → 4.8 (+3.8).
  • Granite 4.0 H 350M: intelligence 1 → 4.8 (+3.8).
  • LFM2 1.2B: intelligence 1 → 4.8 (+3.8).
  • LFM2.5-VL-1.6B: intelligence 1 → 4.8 (+3.8).
  • Llama 3.2 Instruct 1B: intelligence 1 → 4.8 (+3.8).
  • Llama 3 Instruct 8B: intelligence 1 → 4.8 (+3.8).
  • Qwen3 0.6B (Non-reasoning): intelligence 1 → 4.8 (+3.8).
  • Tiny Aya Global: intelligence 1 → 4.8 (+3.8).
  • GLM-5.3 (max): intelligence 48.6 → 44.9 (-3.7).
  • Qwen3 Max Thinking: intelligence 25 → 21.3 (-3.7).
  • MiniMax-M2.1: intelligence 24.6 → 20.9 (-3.7).
  • Claude 4 Opus (Reasoning): intelligence 24.3 → 20.6 (-3.7).
  • Claude 3.5 Sonnet (Oct '24): intelligence 4.2 → 7.9 (+3.7).
  • DeepSeek R1 Distill Llama 70B: intelligence 4.2 → 7.9 (+3.7).
  • DeepSeek R1 Distill Qwen 14B: intelligence 4.1 → 7.8 (+3.7).
  • Falcon-H1R-7B: intelligence 4.1 → 7.8 (+3.7).
  • GPT-4.1 nano: intelligence 4.1 → 7.8 (+3.7).
  • Ling-flash-2.0: intelligence 4.1 → 7.8 (+3.7).
  • Claude Opus 4.7 (Adaptive Reasoning, Max Effort): intelligence 44.3 → 40.7 (-3.6).
  • GPT-5 mini (medium): intelligence 24.2 → 20.6 (-3.6).
  • GPT-5.1 Codex mini (high): intelligence 24 → 20.4 (-3.6).
  • Qwen3.5 Omni Plus: intelligence 24 → 20.4 (-3.6).
  • Gemini 1.5 Pro (Sep '24): intelligence 4.3 → 7.9 (+3.6).
  • Hermes 4 - Llama-3.1 70B (Reasoning): intelligence 4.3 → 7.9 (+3.6).
  • Qwen3 VL 30B A3B Instruct: intelligence 4.3 → 7.9 (+3.6).
  • Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback): intelligence 53.2 → 49.7 (-3.5).
  • GLM-5.2 (max): intelligence 42.1 → 38.6 (-3.5).
  • Grok 4.1 Fast (Reasoning): intelligence 23.9 → 20.4 (-3.5).
  • o3: intelligence 23.7 → 20.2 (-3.5).
  • Qwen2.5 Max: intelligence 4.5 → 8 (+3.5).
  • Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback): intelligence 56.8 → 53.4 (-3.4).
  • Claude Opus 5 (Adaptive Reasoning, Max Effort): intelligence 54.1 → 50.7 (-3.4).
  • Qwen3.8-Flash-Next: intelligence 45.6 → 42.2 (-3.4).
  • K-EXAONE 2.0: intelligence 23.1 → 19.7 (-3.4).
  • Step 3.7 Flash: intelligence 22.9 → 19.5 (-3.4).
  • Mercury 2: intelligence 14.9 → 11.5 (-3.4).
  • DeepSeek R1 0528 Qwen3 8B: intelligence 4.7 → 8.1 (+3.4).
  • Qwen3.5 35B A3B (Reasoning): intelligence 22.6 → 19.3 (-3.3).
  • Claude 4 Sonnet (Reasoning): intelligence 22.2 → 18.9 (-3.3).
  • Gemini 3.1 Flash-Lite: intelligence 19.3 → 16 (-3.3).
  • gpt-oss-120b (high): intelligence 15.6 → 12.3 (-3.3).
  • Devstral 2: intelligence 12.7 → 9.4 (-3.3).
  • Devstral Small 2: intelligence 11.7 → 8.4 (-3.3).
  • EXAONE 4.0 32B (Reasoning): intelligence 4.9 → 8.2 (+3.3).
  • Qwen3 VL 8B (Reasoning): intelligence 4.9 → 8.2 (+3.3).
  • GLM-4.6 (Reasoning): intelligence 21.7 → 18.5 (-3.2).
  • Magistral Small 1: intelligence 5 → 8.2 (+3.2).
  • JT-35B-Flash: intelligence 21.8 → 18.7 (-3.1).
  • KAT-Coder-Pro V1: intelligence 21.7 → 18.6 (-3.1).
  • MiniMax-M2: intelligence 21.7 → 18.6 (-3.1).
  • DeepSeek R1 Distill Qwen 32B: intelligence 5.3 → 8.4 (+3.1).
  • LFM2.5-2.6B: intelligence 5.3 → 8.4 (+3.1).
  • Nanbeige4.1-3B: intelligence 5.3 → 8.4 (+3.1).
  • Qwen3 VL 32B Instruct: intelligence 5.3 → 8.4 (+3.1).
  • Ministral 3 3B: intelligence 1.7 → 4.8 (+3.1).
  • Grok 4 Fast (Reasoning): intelligence 20.8 → 17.9 (-2.9).
  • Magistral Small 1.2: intelligence 5.7 → 8.6 (+2.9).
  • Gemini 2.5 Flash-Lite (Reasoning): intelligence 5.6 → 8.5 (+2.9).
  • Mistral Small 3.1: intelligence 10.3 → 7.4 (-2.9).
  • Gemma 3 27B Instruct: intelligence 2 → 4.9 (+2.9).
  • Claude 3.7 Sonnet (Reasoning): intelligence 20.5 → 17.7 (-2.8).
  • Nemotron 3.5 Lightning: intelligence 16.4 → 13.6 (-2.8).
  • Sonar Reasoning: intelligence 5.9 → 8.7 (+2.8).
  • Gemma 3 12B Instruct: intelligence 1 → 3.8 (+2.8).
  • Claude 3 Opus: intelligence 6 → 8.7 (+2.7).
  • Devstral Small (May '25): intelligence 6 → 8.7 (+2.7).
  • Ling-2.6-1T: intelligence 19.6 → 17 (-2.6).
  • Doubao Seed Code: intelligence 19.5 → 16.9 (-2.6).
  • MiniCPM5-1B (Reasoning): intelligence 6.2 → 8.8 (+2.6).
  • Qwen3 4B 2507 (Reasoning): intelligence 6.2 → 8.8 (+2.6).
  • Sarvam 105B (high): intelligence 6.2 → 8.8 (+2.6).
  • Gemini 2.0 Pro Experimental (Feb '25): intelligence 6.1 → 8.7 (+2.6).
  • Ling 3.0 Flash: intelligence 27.4 → 24.9 (-2.5).
  • Step 3.5 Flash 2603: intelligence 19.5 → 17 (-2.5).
  • Gemma 4 26B A4B (Reasoning): intelligence 19.2 → 16.7 (-2.5).
  • Step 3.5 Flash: intelligence 19.1 → 16.6 (-2.5).
  • GPT-4o (March 2025, chatgpt-4o-latest): intelligence 6.5 → 9 (+2.5).
  • Gemma 4 E4B (Reasoning): intelligence 6.4 → 8.9 (+2.5).
  • Llama 3.3 Nemotron Super 49B v1 (Reasoning): intelligence 6.4 → 8.9 (+2.5).
  • o4-mini (high): intelligence 19.1 → 16.7 (-2.4).
  • DeepSeek V3.2 Exp (Reasoning): intelligence 19 → 16.6 (-2.4).
  • Magistral Medium 1: intelligence 6.7 → 9.1 (+2.4).
  • Solar Pro 2 (Preview) (Reasoning): intelligence 6.7 → 9.1 (+2.4).
  • Devstral Medium: intelligence 6.6 → 9 (+2.4).
  • Llama Nemotron Super 49B v1.5 (Reasoning): intelligence 6.6 → 9 (+2.4).
  • Tri-21B-Think: intelligence 6.6 → 9 (+2.4).
  • Claude 3.5 Haiku: intelligence 6.5 → 8.9 (+2.4).
  • Gemini 2.0 Flash (Feb '25): intelligence 6.5 → 8.9 (+2.4).
  • Gemini 2.5 Pro: intelligence 19 → 16.7 (-2.3).
  • Ling-1T: intelligence 6.9 → 9.2 (+2.3).
  • Nova Premier: intelligence 6.9 → 9.2 (+2.3).
  • Mistral Medium 3: intelligence 6.7 → 9 (+2.3).
  • Trinity Large Thinking: intelligence 13.1 → 10.9 (-2.2).
  • Motif-2-12.7B-Reasoning: intelligence 7 → 9.2 (+2.2).
  • Qwen3 Max: intelligence 17.6 → 15.6 (-2.0).
  • Qwen3 VL 30B A3B (Reasoning): intelligence 7.5 → 9.5 (+2.0).
  • QwQ 32B: intelligence 7.5 → 9.5 (+2.0).
  • Gemini 2.0 Flash Thinking Experimental (Jan '25): intelligence 7.4 → 9.4 (+2.0).
  • Mistral Small 3.2: intelligence 5 → 7 (+2.0).
  • Ministral 3 8B: intelligence 3.5 → 5.5 (+2.0).
  • GPT-6 Astra (max): intelligence 54.7 → 52.8 (-1.9).
  • Gemini 2.5 Flash Preview (Sep '25) (Reasoning): intelligence 17.4 → 15.5 (-1.9).
  • Kimi K2 0905: intelligence 17.2 → 15.3 (-1.9).
  • o1: intelligence 17.1 → 15.2 (-1.9).
  • GPT-4.5 (Preview): intelligence 7.7 → 9.6 (+1.9).
  • Qwen3 Coder 30B A3B Instruct: intelligence 7.7 → 9.6 (+1.9).
  • Tri-21B-think Preview: intelligence 7.7 → 9.6 (+1.9).
  • DiffusionGemma 26B A4B: intelligence 7.6 → 9.5 (+1.9).
  • Qwen3 235B A22B (Reasoning): intelligence 7.6 → 9.5 (+1.9).
  • DeepSeek V3 (Dec '24): intelligence 10.4 → 8.5 (-1.9).
  • Llama 4 Scout: intelligence 4.6 → 6.5 (+1.9).
  • Ling 2.6 Flash: intelligence 7.9 → 9.7 (+1.8).
  • Qwen3 Next 80B A3B Instruct: intelligence 7.8 → 9.6 (+1.8).
  • Gemini 2.5 Pro Preview (Mar' 25): intelligence 16.7 → 15 (-1.7).
  • GLM-4.7-Flash (Reasoning): intelligence 16.6 → 14.9 (-1.7).
  • Granite 4.2 30B: intelligence 16.5 → 14.8 (-1.7).
  • Command A+: intelligence 15.6 → 13.9 (-1.7).
  • o1-mini: intelligence 8.1 → 9.8 (+1.7).
  • Grok 3 mini Reasoning (high): intelligence 16.2 → 14.6 (-1.6).
  • K2-V2 (high): intelligence 8.3 → 9.9 (+1.6).
  • Granite 4.2 3B: intelligence 10.7 → 9.1 (-1.6).
  • Solar Pro 3: intelligence 9.4 → 7.8 (-1.6).
  • DeepSeek V3.2 Speciale: intelligence 16 → 14.5 (-1.5).
  • Granite 4.2 8B: intelligence 13.9 → 12.4 (-1.5).
  • MiniMax M1 40k: intelligence 8.5 → 10 (+1.5).
  • Mistral Medium 3.1: intelligence 8.4 → 9.9 (+1.5).
  • Qwen3 VL 235B A22B Instruct: intelligence 8.4 → 9.9 (+1.5).
  • K-EXAONE (Reasoning): intelligence 15.8 → 14.4 (-1.4).
  • ERNIE 5.0 Thinking Preview: intelligence 15.7 → 14.3 (-1.4).
  • Gemma 4 12B (Reasoning): intelligence 15.6 → 14.2 (-1.4).
  • GPT-4.1 mini: intelligence 8.8 → 10.2 (+1.4).
  • Mistral Large 3: intelligence 11.1 → 9.7 (-1.4).
  • Nova 2.0 Pro Preview (medium): intelligence 15.5 → 14.2 (-1.3).
  • Grok Code Fast 1: intelligence 15.4 → 14.1 (-1.3).
  • Mistral Small 4 (Reasoning): intelligence 12.8 → 11.5 (-1.3).
  • Nemotron 3 Nano Omni 30B A3B Reasoning: intelligence 9 → 10.3 (+1.3).
  • Celeris-1: intelligence 7.6 → 6.3 (-1.3).
  • Apriel-v1.5-15B-Thinker: intelligence 15 → 13.8 (-1.2).
  • Gemini 2.5 Flash-Lite Preview (Sep '25) (Reasoning): intelligence 9.2 → 10.4 (+1.2).
  • Grok 3 Reasoning Beta: intelligence 9.2 → 10.4 (+1.2).
  • Solar Open 100B (Reasoning): intelligence 9.2 → 10.4 (+1.2).
  • Qwen3 30B A3B 2507 (Reasoning): intelligence 8.6 → 9.8 (+1.2).
  • DeepSeek V3.1 (Non-reasoning): intelligence 14.8 → 13.7 (-1.1).
  • Qwen3.5 9B (Reasoning): intelligence 14.8 → 13.7 (-1.1).
  • Nova 2.0 Omni (medium): intelligence 14.7 → 13.6 (-1.1).
  • Qwen3 VL 235B A22B (Reasoning): intelligence 14.4 → 13.4 (-1.0).
  • INTELLECT-3: intelligence 9.6 → 10.6 (+1.0).
  • Llama 4 Maverick: intelligence 10.3 → 9.3 (-1.0).
  • Apriel-v1.6-15B-Thinker: intelligence 14.3 → 13.4 (-0.9).
  • Nova 2.0 Lite (high): intelligence 14.3 → 13.4 (-0.9).
  • EXAONE 4.5 33B: intelligence 14 → 13.2 (-0.8).
  • DeepSeek R1 0528 (May '25): intelligence 13.9 → 13.1 (-0.8).
  • Gemini 2.5 Flash (Reasoning): intelligence 13.9 → 13.1 (-0.8).
  • Qwen3.5 4B (Reasoning): intelligence 13.9 → 13.1 (-0.8).
  • MiniCPM5-2B: intelligence 15 → 14.3 (-0.7).
  • GPT-5 nano (high): intelligence 13.7 → 13 (-0.7).
  • North Mini Code: intelligence 13.5 → 12.8 (-0.7).
  • Ring-1T: intelligence 10.2 → 10.9 (+0.7).
  • G9v3-3B: intelligence 10.1 → 10.8 (+0.7).
  • Kimi K2: intelligence 13.3 → 12.7 (-0.6).
  • DeepSeek R1 (Jan '25): intelligence 12 → 11.4 (-0.6).
  • GLM-4.5-Air: intelligence 10.5 → 11.1 (+0.6).
  • Mi:dm K 2.5 Pro: intelligence 10.4 → 11 (+0.6).
  • GLM-4.5 (Reasoning): intelligence 13.3 → 12.8 (-0.5).
  • GPT-4.1: intelligence 13.2 → 12.7 (-0.5).
  • GLM-4.6V (Reasoning): intelligence 10.7 → 11.2 (+0.5).
  • Qwen3 Next 80B A3B (Reasoning): intelligence 10.7 → 11.2 (+0.5).
  • DeepSeek V3 0324: intelligence 9.2 → 9.7 (+0.5).
  • Ministral 3 14B: intelligence 5.5 → 6 (+0.5).

Ny i HF-trending

  • deepseek-ai/DeepSeek-V4.1-Flash er ny i Hugging Faces trending (score 986).
  • nex-agi/Nex-N2.5-mini er ny i Hugging Faces trending (score 477).
  • nex-agi/Nex-N2.5-Pro er ny i Hugging Faces trending (score 439).
  • WarmBloodAban/Minimax-h3_Singularity er ny i Hugging Faces trending (score 247).
  • m-a-p/YuE2-3B er ny i Hugging Faces trending (score 66).
  • TokenRhythm/NeoHorse-1-4B er ny i Hugging Faces trending (score 65.5).
  • inclusionAI/Ling-3.0-flash-VL er ny i Hugging Faces trending (score 64).
  • tencent/AuK er ny i Hugging Faces trending (score 52).
  • Jackrong/Qwopus3.8-27B-Flash er ny i Hugging Faces trending (score 44).
  • pyannote/segmentation-3.0 er ny i Hugging Faces trending (score 36).
  • Edge0/Edge0-35B-A3B-preview er ny i Hugging Faces trending (score 34).
  • nex-agi/Nex-N2.5-Max er ny i Hugging Faces trending (score 34).
  • thesysdev/OUI-1 er ny i Hugging Faces trending (score 34).

Intelligence Index — Artificial Analysis

#ModelIntel.Agentic CreatorUdgivet
1 Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) 5 effort-varianter 53.4 58 Anthropic 2026-09-01
2 GPT-6 Astra (max) 6 effort-varianter 52.8 51.5 OpenAI 2026-09-03
3 Claude Opus 5 (Adaptive Reasoning, Max Effort) 5 effort-varianter 50.7 56.2 Anthropic 2026-07-24
4 Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) 49.7 51 Anthropic 2026-06-09
5 Muse Spark 1.3 (max) 2 effort-varianter 48.2 55.7 Meta 2026-09-02
6 GPT-5.6 Sol (max) 6 effort-varianter 47.1 50.5 OpenAI 2026-07-09
7 GLM-5.3 (max) ⬇ hf.co/zai-org/GLM-5.3 44.9 53.4 Z AI 2026-08-18
8 Grok 4.6 (high) 4 effort-varianter 44.4 53.4 SpaceXAI 2026-08-12
9 Kimi K3 (max) 2 effort-varianter ⬇ hf.co/moonshotai/Kimi-K3 43.8 50.6 Kimi 2026-07-16
10 GPT-5.6 Terra (max) 6 effort-varianter 42.3 43.7 OpenAI 2026-07-09
11 Qwen3.8-Flash-Next ⬇ hf.co/Qwen/Qwen3.8-Flash-Next 42.2 Alibaba 2026-08-26
12 Claude Opus 4.8 (Adaptive Reasoning, Max Effort) 42 42.6 Anthropic 2026-05-28
13 GLM-5.3-Flash ⬇ hf.co/zai-org/GLM-5.3-Flash 41.9 51.2 Z AI 2026-08-26
14 Gemini 3.8 Flash (high) 3 effort-varianter 41.2 41.1 Google 2026-09-02
15 Claude Opus 4.7 (Adaptive Reasoning, Max Effort) 2 effort-varianter 40.7 39.5 Anthropic 2026-04-16
16 Qwen3.8 Max 40.3 49.6 Alibaba 2026-08-03
17 Qwen3.8 2.4T A95B ⬇ hf.co/Qwen/Qwen3.8-2.4T-A95B 40 50.4 Alibaba 2026-08-12
18 Muse Spark 1.2 (xhigh) 39.8 44 Meta 2026-08-05
19 Gemini 3.7 Flash (medium) 3 effort-varianter 39.6 Google 2026-08-13
20 Grok 4.5 (high) 39.1 42.1 SpaceXAI 2026-07-08
21 GPT-5.4 (xhigh) 3 effort-varianter 39 OpenAI 2026-03-05
22 GLM-5.2 (max) 2 effort-varianter ⬇ hf.co/zai-org/GLM-5.2 38.6 39.4 Z AI 2026-06-16
23 GPT-5.5 (xhigh) 5 effort-varianter 38.6 37.3 OpenAI 2026-04-23
24 Claude Sonnet 5 (Adaptive Reasoning, Max Effort) 6 effort-varianter 38.4 44.3 Anthropic 2026-06-30
25 GPT-5.6 Luna (max) 6 effort-varianter 37.5 42.7 OpenAI 2026-07-09
26 DeepSeek V4 Pro 0813 (Reasoning, Max Effort) ⬇ hf.co/deepseek-ai/DeepSeek-V4-Pro 36.3 42.3 DeepSeek 2026-08-13
27 Agnes 2.5 Pro Beta 35.2 Sapiens AI 2026-08-26
28 DeepSeek V4 Flash Vision (Reasoning, Max Effort) 35 48.1 DeepSeek 2026-08-21
29 DeepSeek V4 Flash 0731 (Reasoning, Max Effort) ⬇ hf.co/deepseek-ai/DeepSeek-V4-Flash-0731 34.5 41.7 DeepSeek 2026-07-31
30 Gemini 3.6 Flash (high) 34.3 30.2 Google 2026-07-21
31 Muse Spark 1.1 (xhigh) 34.3 27.5 Meta 2026-07-09
32 K2 Horizon 375B A23B ⬇ hf.co/IFM/K2-Horizon-375B-A23B 33.9 40.9 MBZUAI Institute of Foundation Models 2026-09-03
33 Qwen3.8 27B (xhigh) 4 effort-varianter ⬇ hf.co/Qwen/Qwen3.8-27B 33.9 46.5 Alibaba 2026-08-14
34 Gemini 3.5 Flash (medium) 3 effort-varianter 33.6 Google 2026-05-19
35 Motif 3 2 effort-varianter ⬇ hf.co/Motif-Technologies/Motif-3 33.6 Motif Technologies 2026-08-12
36 GPT-5.3 Codex (xhigh) 32.5 OpenAI 2026-02-05
37 Claude Opus 4.6 (Adaptive Reasoning, Max Effort) 2 effort-varianter 31.9 Anthropic 2026-02-05
38 Kimi K2.6 2 effort-varianter 31.3 22.1 Kimi 2026-04-20
39 Muse Spark 31.3 Meta 2026-04-08
40 DeepSeek V4 Pro (Reasoning, Max Effort) 3 effort-varianter ⬇ hf.co/deepseek-ai/DeepSeek-V4-Pro 30.9 27.7 DeepSeek 2026-04-24

Hugging Face — trending familier

#Rod-repoTrendingLikes DownloadsParamsAlder
1 deepseek-ai/DeepSeek-V4.1-Flash 986 1011 6 763.2B 0 d
2 openbmb/MiniCPM5-2B +1 derivater 870 1055 42289 2.5B 4 d
3 XHToken/Spark-X2.5-4B-Base +2 derivater 729 1041 15930 4.1B 17 d
4 Qwen/Qwen3.8-27B +16 derivater 530 14589 7322476 27.8B 36 d
5 nex-agi/Nex-N2.5-mini 477 629 2444 35.1B 2 d
6 nex-agi/Nex-N2.5-Pro 439 580 3 2 d
7 Lightricks/LTX-2.5 333 3333 1740572 49 d
8 google/timesfm-3.0-pytorch 326 704 483787 330.7M 16 d
9 WarmBloodAban/Minimax-h3_Singularity 247 255 96682 4 d
10 IFM/K2-Horizon-MoVA-36B-A4B +1 derivater 245 267 4488 37.4B 8 d
11 Qwen/Qwen3.8-Flash-Next +4 derivater 231 5061 564079 180.0B 17 d
12 deepseek-ai/DeepSeek-V4-Flash-Vision-Exp 230 855 400892 304.6B 10 d
13 MiniMaxAI/MiniMax-H3 +8 derivater 207 5111 5080204 33.1B 44 d
14 zai-org/GLM-5.3-Flash +3 derivater 183 2215 1023103 321.3B 16 d
15 zai-org/GLM-5.3 +1 derivater 163 1797 552019 753.3B 16 d

Params er safetensors.total, altså lagrede vægte inklusive kvantiserede tensorer — ikke et mål for hvad modellen kræver af hardware.