Ctrl K

Models evolve.
Stay current.

Compare performance, cost, and capabilities at a glance.

Compare ModelsExplore Radar
Claude Opus 5.557.6
GPT-6 Astra52.7
Intelligence IndexEval. cost / task (USD) →
168models in the catalog83of them independently ranked129measured configurations
42model developers

What’s happening in AI

All news
Meta / NVIDIA / ReflectionOpen models3 min read

Meta, NVIDIA and Reflection release open weights for very different machines

Meta demonstrates an 816B model on two Mac Studios, NVIDIA targets a 24GB weight footprint, and Reflection releases Beam under Apache 2.0.

Reported 7 sourcesRead the story

Coming up

All events
  1. In 2 daysNVIDIA GTC Berlin 2026STATION-Berlin · Keynote at Tempodrom, Berlin20 – 22 October 2026
  2. In 30 daysMicrosoft Ignite 2026Moscone Center, San Francisco · Online17 – 20 November 2026
  1. GoogleModel accessGoogle opens Argon to developers and paid Gemini users
  2. AnthropicAI agentsClaude AI helped map the sky—and sent a fake police tip
  3. OpenAIAI securityOpenAI says a 15,000-user network tried to extract hidden AI reasoning

Model landscape right now

Explore benchmarks

Overall

points

Intelligence Index

  1. 01Claude Opus 5.5Max57.6
  2. 02Claude Sonnet 5.5Max56.0
  3. 03Muse Wonder FullMax55.8
  4. 04Claude Fable 5.1Max53.4
  5. 05GPT-6 AstraMax52.7

Reasoning

%

GPQA Diamond

  1. 01GPT-6 AstraMax96.1
  2. 02Gemini 3.8 FlashHigh95.3
  3. 03Gemini 3.1 Pro PreviewReasoning94.1
  4. 04Claude Fable 5.1Max93.7
  5. 05Qwen3.8 2.4T A95BReasoning93.5

Coding

%

SciCode

  1. 01Claude Opus 5.5Max66.9
  2. 02Claude Fable 5.1Max63.1
  3. 03Muse Wonder FullMax62.6
  4. 04Gemini 4 ArgonHigh61.8
  5. 05Claude Sonnet 5.5Max61.0

Efficiency

pts/$

Intelligence Index 40+

  1. 01MiMo-V2.6-ProReasoning347.7
  2. 02Claude Haiku 5.5Max204.0
  3. 03GLM-5.3-FlashMax147.4
  4. 04GPT-6.1 SolMax71.6
  5. 05Step 5 PreviewReasoning42.6

Artificial Analysis measurements, each model at its highest-scoring measured setting. Scores cover defined tasks, not every workload; ties share a rank, and gaps don’t establish statistical significance. Methodology

How much does intelligence cost?

Explore trade-offs
Intelligence Index Cost frontier
0102030405060$0.001$0.01$0.10$1$10Claude Opus 5.5, Max, Default Fallback, 57.6 intelligence, $5.98 per evaluation taskClaude Sonnet 5.5, Max, Default Fallback, 56 intelligence, $5.46 per evaluation taskClaude Fable 5.1, Max, Default Fallback, 53.4 intelligence, $7.63 per evaluation taskGPT-6 Astra, Max, 52.7 intelligence, $3.26 per evaluation taskGemini 4 Argon, High, 52.6 intelligence, $1.99 per evaluation taskMuse Wonder, Max, 52.4 intelligence, $4.52 per evaluation taskGPT-6.1 Sol, Max, 51.8 intelligence, $0.72 per evaluation taskMuse Spark 1.3, Max, 48.1 intelligence, $1.60 per evaluation taskGrok 4.7, Xhigh, 46.4 intelligence, $3.74 per evaluation taskMiMo-V2.6-Pro, Reasoning, 46.3 intelligence, $0.13 per evaluation taskQwen3.8 Max (0902), 0902, 45.4 intelligence, $5.41 per evaluation taskGLM-5.3, Max, 44.8 intelligence, $2.01 per evaluation taskStep 5 Preview, Reasoning, 43.7 intelligence, $1.03 per evaluation taskKimi-K3, Max, 43.6 intelligence, $2.00 per evaluation taskClaude Haiku 5.5, Max, 43.4 intelligence, $0.21 per evaluation taskGPT-5.6 Terra, Max, 42.1 intelligence, $1.40 per evaluation taskGLM-5.3-Flash, Max, 41.8 intelligence, $0.28 per evaluation taskLing-3.1-flash, Reasoning, 41.1 intelligence, $1.00 per evaluation taskGemini 3.8 Flash, High, 40.9 intelligence, $1.24 per evaluation taskQwen3.8 2.4T A95B, Reasoning, 39.9 intelligence, $2.16 per evaluation taskQwen3.8-Flash-Next, Reasoning, 39.8 intelligence, $0.37 per evaluation taskDeepSeek-V4.1-Flash, Max, 39.5 intelligence, $0.27 per evaluation taskMuse Spark Open, Xhigh, 38.9 intelligence, $0.84 per evaluation taskMistral Large 4, Reasoning, 38.4 intelligence, $1.13 per evaluation taskGPT-6 Luna, Max, 38.1 intelligence, $0.07 per evaluation taskMiMo-V2.6-Flash, Reasoning, 37.9 intelligence, $0.06 per evaluation taskDeepSeek-V4-Pro-0813, Max, 36 intelligence, $0.67 per evaluation taskDeepSeek V4 Flash Vision, Max, 34.8 intelligence, $0.31 per evaluation taskQwen3.8-27B, Xhigh, 33.7 intelligence, $1.01 per evaluation taskBeam 1.0, Reasoning, 30.3 intelligence, $0.57 per evaluation taskGemini 3.1 Pro Preview, Reasoning, 29.7 intelligence, $1.30 per evaluation taskMiniMax-M3, Reasoning, 29.2 intelligence, $0.51 per evaluation taskQuasar 438B, Max, Based on GLM-5.2, 26.7 intelligence, $2.02 per evaluation taskNemotron 3.5 Super VL, Reasoning, 26.5 intelligence, $0.19 per evaluation taskApodex 1.1, Reasoning, 26.4 intelligence, $0.46 per evaluation taskGPT-5.5 Instant (June 2026), June 2026, 26 intelligence, $0.69 per evaluation taskKimi K2.7 Code, Reasoning, 25.8 intelligence, $0.54 per evaluation taskInkling Small, Reasoning, 25.7 intelligence, $0.09 per evaluation taskHy3, Reasoning, 25.3 intelligence, $0.07 per evaluation taskQwen3.7 Plus, Reasoning, 25.2 intelligence, $0.22 per evaluation taskInkling, Extra high, 25 intelligence, $0.61 per evaluation taskNemotron 3 Ultra 550B A55B, Reasoning, 22.9 intelligence, $0.58 per evaluation taskGemini 3.5 Flash-Lite, High, 22.2 intelligence, $0.19 per evaluation taskLongCat 2.0, Reasoning, 19.1 intelligence, $0.06 per evaluation taskQwen3.5 397B A17B, Reasoning, 18.4 intelligence, $0.47 per evaluation taskQwen3.6 35B A3B, Reasoning, 18.2 intelligence, $0.48 per evaluation taskMuse Glimmer, High, 17.5 intelligence, $0.06 per evaluation taskRing-2.6-1T, Extra high, 16.6 intelligence, $0.29 per evaluation taskQwen3.5 122B A10B, Reasoning, 15.6 intelligence, $0.32 per evaluation taskMistral Medium 3.5, High, 14.2 intelligence, $0.50 per evaluation taskNemotron 3.5 Lightning, Reasoning, 12.9 intelligence, $0.09 per evaluation taskGranite 4.2 30B, Reasoning, 12.7 intelligence, $0.07 per evaluation taskgpt-oss-120b, High, 11.6 intelligence, $0.09 per evaluation taskMistral Small 4, Reasoning, 11.3 intelligence, $0.02 per evaluation taskQwen3.5 9B, Reasoning, 11.2 intelligence, $0.21 per evaluation taskGranite 4.2 8B, Reasoning, 11.1 intelligence, $0.02 per evaluation taskTrinity Large Thinking, Reasoning, 10.8 intelligence, $0.12 per evaluation taskLlama 4 Maverick, Standard, 9.3 intelligence, $0.04 per evaluation taskQwen3 Coder Next, Standard, 9.2 intelligence, $0.47 per evaluation taskGranite 4.2 3B, Reasoning, 9.1 intelligence, $0.01 per evaluation taskgpt-oss-20b, High, 9 intelligence, $0.01 per evaluation taskNVIDIA Nemotron 3 Nano 30B A3B, Reasoning, 8.9 intelligence, $0.02 per evaluation taskLlama 4 Scout, Standard, 6.5 intelligence, $0.11 per evaluation taskCeleris-1, Standard, 6.3 intelligence, $0.05 per evaluation taskMinistral 3 14B, Standard, 6 intelligence, $0.02 per evaluation taskMinistral 3 8B, Standard, 5.5 intelligence, $0.01 per evaluation taskMinistral 3 3B, Standard, 4.8 intelligence, $0.01 per evaluation taskClaude Opus 5.5MiMo-V2.6-Pro

Evaluation cost per task (USD, log scale)

Latest ranked releases

All models

  1. Beam 1.0

    Overall rank
    #33 / 83
    Index
    30.3
    USD / task
    $0.57

  2. Muse Spark Open

    Overall rank
    #24 / 83
    Index
    38.9
    USD / task
    $0.84

  3. Nemotron 3.5 Bolt 74B-A9B

    Overall rank
    #51 / 83
    Index
    21.4
    USD / task
    Not reported

  4. Nemotron 3.5 Super VL

    Overall rank
    #37 / 83
    Index
    26.5
    USD / task
    $0.19

  5. Muse Wonder

    Overall rank
    #7 / 83
    Index
    52.4
    USD / task
    $4.52

  6. Muse Wonder Full

    Overall rank
    #3 / 83
    Index
    55.8
    USD / task
    Not reported

Head to head

Choose your own
The top two developersIntelligence Index 57.6 versus 55.8; evaluation cost per task $5.98 versus Not reported.1.8 index points apartFull comparison Capability versus costIntelligence Index 57.6 versus 46.3; evaluation cost per task $5.98 versus $0.13.11.3 index points apartFull comparison Open versus closed weightsIntelligence Index 46.3 versus 46.4; evaluation cost per task $0.13 versus $3.74.0.1 index points apartFull comparison