Categories

Reference table

AI model ranking

Compare leading AI models by benchmark score, provider, API price, and output speed.

Benchmark: Artificial Analysis Intelligence Index, Updated

Methodology

Top score

Claude Opus 5.5 (Max, Default Fallback)

57.6

Lowest listed price

K2 Horizon 375B A23B

$0.000 / 1M tokens

Fastest

K2 Horizon 375B A23B

891.4 tok/s

Intelligence ranking

The 15 highest scores on the Intelligence Index. A longer bar is a stronger model; the badge is the reasoning effort the score was measured at.

  1. 1
    Claude Opus 5.5Max
    57.6
  2. 2
    Claude Sonnet 5.5Max
    56.0
  3. 3
    Claude Opus 5.5XHigh
    56.0
  4. 4
    Muse Wonder FullMax
    55.8
  5. 5
    Claude Opus 5.5High
    53.6
  6. 6
    Claude Fable 5.1Max
    53.4
  7. 7
    Claude Fable 5.1XHigh
    53.2
  8. 8
    GPT-6 AstraMax
    52.7
  9. 9
    Gemini 4 ArgonHigh
    52.6
  10. 10
    Muse WonderMax
    52.4
  11. 11
    GPT-6 AstraXHigh
    52.4
  12. 12
    Claude Sonnet 5.5XHigh
    51.9
  13. 13
    GPT-6.1 SolMax
    51.8
  14. 14
    Claude Opus 5.5Medium
    51.2
  15. 15
    Claude Fable 5.1High
    51.2

Benchmark: Artificial Analysis Intelligence Index, Updated

Intelligence vs cost per task

Each dot is a model at one effort. The line joins the models that score higher than everything cheaper: the best buy at each price.

  • Best score for its cost (the frontier)
  • Other models
2030405060$0.01$0.03$0.1$0.3$1$3$10Cost per task, USD (log scale, cheaper to the left)Intelligence IndexGPT-6 Luna MediumGPT-6 Luna HighGPT-6 Luna XHighMiMo-V2.6-FlashGPT-6 Luna MaxClaude Haiku 5.5 XHighGPT-6.1 Sol LowMiMo-V2.6-ProGPT-6.1 Sol MediumGPT-6.1 Sol HighGPT-6.1 Sol XHighGPT-6.1 Sol MaxClaude Opus 5.5 HighClaude Opus 5.5 XHighClaude Sonnet 5.5 MaxClaude Opus 5.5 Max

Benchmark: Artificial Analysis Intelligence Index, Updated

Cost per task is what it costs to run one task of the Intelligence Index.

Fastest models

Median output speed in tokens per second, the 10 fastest.

  1. 1
    K2 Horizon 375B A23B

    Institute of Foundation Models · Index 30.5

    891
  2. 2
    Muse Spark 1.3XHigh

    Meta · Index 45.1

    283
  3. 3
    Claude Haiku 5.5Max

    Anthropic · Index 43.4

    238
  4. 4
    Ling 3.1 Flash

    InclusionAI · Index 41.1

    213
  5. 5
    Claude Haiku 5.5XHigh

    Anthropic · Index 41.2

    181
  6. 6
    Claude Haiku 5.5High

    Anthropic · Index 37.8

    177
  7. 7
    DeepSeek V4.1 FlashMax

    DeepSeek · Index 39.5

    177
  8. 8
    DeepSeek V4 Flash VisionMax

    DeepSeek · Index 34.8

    173
  9. 9
    Claude Haiku 5.5Low

    Anthropic · Index 29.5

    161
  10. 10
    Muse WonderMax

    Meta · Index 52.4

    161

Benchmark: Artificial Analysis Intelligence Index, Updated

Speed is measured by Artificial Analysis.

Cost per task

The 10 cheapest models to run one Intelligence Index task, split into what the model read, thought and wrote.

  • Input
  • Reasoning
  • Answer
  1. 1
    GPT-6 LunaMedium

    OpenAI · Index 29.9

    $0.017
  2. 2
    Claude Haiku 5.5Low

    Anthropic · Index 29.5

    $0.024
  3. 3
    GPT-6 LunaHigh

    OpenAI · Index 32.9

    $0.029
  4. 4
    GPT-6 LunaXHigh

    OpenAI · Index 34.6

    $0.042
  5. 5
    Claude Haiku 5.5Medium

    Anthropic · Index 34.5

    $0.047
  6. 6
    MiMo-V2.6-Flash

    Xiaomi · Index 37.9

    $0.062
  7. 7
    GPT-6 LunaMax

    OpenAI · Index 38.1

    $0.068
  8. 8
    Claude Haiku 5.5High

    Anthropic · Index 37.8

    $0.079
  9. 9
    Claude Haiku 5.5XHigh

    Anthropic · Index 41.2

    $0.12
  10. 10
    GPT-6.1 SolLow

    OpenAI · Index 42.1

    $0.13

Benchmark: Artificial Analysis Intelligence Index, Updated

Cheapest first; the Index beside each name shows what it buys.

Benchmark board

Model ranking

Rank, model, provider and reasoning effort, Artificial Analysis Intelligence Index, blended API price, cost per task, and output speed.

RankModelProviderIndexBlended priceCost per taskSpeed
#1Claude Opus 5.5 (Max, Default Fallback)
AnthropicMax effort
57.6$2.940/M$5.9895.7 tok/s
#2Claude Sonnet 5.5 (Max, Default Fallback)
AnthropicMax effort
56.0$1.470/M$5.46132.7 tok/s
#3Claude Opus 5.5 (Xhigh, Default Fallback)
AnthropicXHigh effort
56.0$2.940/M$3.4676.0 tok/s
#4Muse Wonder Full (Max)
MetaMax effort
55.8Not listedNot listedNot listed
#5Claude Opus 5.5 (High, Default Fallback)
AnthropicHigh effort
53.6$2.940/M$1.8277.1 tok/s
#6Claude Fable 5.1 (Max, Default Fallback)
AnthropicMax effort
53.4$7.175/M$7.6371.8 tok/s
#7Claude Fable 5.1 (Xhigh, Default Fallback)
AnthropicXHigh effort
53.2$7.175/M$5.9863.3 tok/s
#8GPT-6 Astra (Max)
OpenAIMax effort
52.7$7.700/M$3.2644.2 tok/s
#9Gemini 4 Argon (High)
GoogleHigh effort
52.6$1.470/M$1.99Not listed
#10Muse Wonder (Max)
MetaMax effort
52.4$2.940/M$4.52160.5 tok/s
#11GPT-6 Astra (Xhigh)
OpenAIXHigh effort
52.4$7.700/M$2.3144.3 tok/s
#12Claude Sonnet 5.5 (Xhigh, Default Fallback)
AnthropicXHigh effort
51.9$1.470/M$2.01100.2 tok/s
#13GPT-6.1 Sol (Max)
OpenAIMax effort
51.8$1.470/M$0.7248.5 tok/s
#14Claude Opus 5.5 (Medium, Default Fallback)
AnthropicMedium effort
51.2$2.940/M$1.3484.8 tok/s
#15Claude Fable 5.1 (High, Default Fallback)
AnthropicHigh effort
51.2$7.175/M$3.9155.2 tok/s
#16GPT-6.1 Sol (Xhigh)
OpenAIXHigh effort
51.0$1.470/M$0.3951.1 tok/s
#17GPT-6 Astra (High)
OpenAIHigh effort
50.9$7.700/M$1.7344.7 tok/s
#18GPT-6.1 Sol (High)
OpenAIHigh effort
50.2$1.470/M$0.3250.6 tok/s
#19GPT-6 Astra (Medium)
OpenAIMedium effort
49.6$7.700/M$1.5446.2 tok/s
#20Claude Fable 5.1 (Medium, Default Fallback)
AnthropicMedium effort
48.9$7.175/M$2.9853.9 tok/s
#21Muse Spark 1.3 (Max)
MetaMax effort
48.1$0.780/M$1.60127.7 tok/s
#22GPT-6.1 Sol (Medium)
OpenAIMedium effort
47.8$1.470/M$0.2146.4 tok/s
#23Claude Fable 5.1 (Low, Default Fallback)
AnthropicLow effort
46.8$7.175/M$2.3756.8 tok/s
#24Claude Sonnet 5.5 (High, Default Fallback)
AnthropicHigh effort
46.8$1.470/M$0.88102.7 tok/s
#25Grok 4.7 (Xhigh)
SpaceXAIXHigh effort
46.4$1.350/M$3.7479.5 tok/s
#26Grok 4.7 (High)
SpaceXAIHigh effort
46.3$1.350/M$2.7376.5 tok/s
#27MiMo-V2.6-Pro
Xiaomi
46.3$0.177/M$0.1347.3 tok/s
#28GPT-6 Astra (Low)
OpenAILow effort
45.8$7.700/M$0.8242.3 tok/s
#29Qwen3.8 Max (0902)
Qwen
45.4$1.175/M$5.4139.8 tok/s
#30Muse Spark 1.3 (Xhigh)
MetaXHigh effort
45.1$0.780/M$1.37282.8 tok/s
#31GLM-5.3 (Max)
Z.aiMax effort
44.8$0.902/M$2.0162.4 tok/s
#32Step 5 Preview
StepFun
43.7$0.540/M$1.0390.8 tok/s
#33Kimi K3 (Max)
KimiMax effort
43.6$2.310/M$2.0043.3 tok/s
#34Claude Haiku 5.5 (Max)
AnthropicMax effort
43.4$0.077/M$0.21238.2 tok/s
#35Claude Opus 5.5 (Low, Default Fallback)
AnthropicLow effort
42.3$2.940/M$0.5583.5 tok/s
#36Grok 4.7 (Low)
SpaceXAILow effort
42.2$1.350/M$1.2563.7 tok/s
#37GPT-6.1 Sol (Low)
OpenAILow effort
42.1$1.470/M$0.1344.3 tok/s
#38GPT-5.6 Terra (Max)
OpenAIMax effort
42.1$1.740/M$1.40120.7 tok/s
#39GLM-5.3-Flash
Z.aiMax effort
41.8$0.101/M$0.2864.5 tok/s
#40Claude Haiku 5.5 (Xhigh)
AnthropicXHigh effort
41.2$0.077/M$0.12181.1 tok/s
#41Ling 3.1 Flash
InclusionAI
41.1$0.192/M$1.00212.6 tok/s
#42Gemini 3.8 Flash (High)
GoogleHigh effort
40.9$0.578/M$1.24144.9 tok/s
#43Claude Sonnet 5.5 (Medium, Default Fallback)
AnthropicMedium effort
40.8$1.470/M$0.48105.3 tok/s
#44Qwen3.8 2.4T A95B
Qwen
39.9$1.175/M$2.1639.1 tok/s
#45Qwen3.8-Flash-Next
Qwen
39.8$0.088/M$0.3752.0 tok/s
#46Gemini 3.8 Flash (Medium)
GoogleMedium effort
39.8$0.578/M$0.93Not listed
#47DeepSeek V4.1 Flash (Max)
DeepSeekMax effort
39.5$0.184/M$0.27176.6 tok/s
#48Mistral Large 4 Preview
Mistral
38.4$0.788/M$1.13Not listed
#49GPT-6 Luna (Max)
OpenAIMax effort
38.1$0.077/M$0.068120.4 tok/s
#50GPT-5.6 Terra (Xhigh)
OpenAIXHigh effort
38.0$1.740/M$0.6398.4 tok/s
#51MiMo-V2.6-Flash
Xiaomi
37.9$0.058/M$0.06253.8 tok/s
#52Claude Haiku 5.5 (High)
AnthropicHigh effort
37.8$0.077/M$0.079177.1 tok/s
#53DeepSeek V4 Pro 0813 (Max)
DeepSeekMax effort
36.0$0.691/M$0.6792.0 tok/s
#54Claude Sonnet 5.5 (Low, Default Fallback)
AnthropicLow effort
35.9$1.470/M$0.3598.9 tok/s
#55Agnes 2.5 Pro Beta
Sapiens AI
35.2$0.057/MNot listed119.4 tok/s
#56DeepSeek V4 Flash Vision (Max)
DeepSeekMax effort
34.8$0.230/M$0.31173.0 tok/s
#57GPT-6 Luna (Xhigh)
OpenAIXHigh effort
34.6$0.077/M$0.042113.2 tok/s
#58Claude Haiku 5.5 (Medium)
AnthropicMedium effort
34.5$0.077/M$0.047150.9 tok/s
#59GLM-5.3 (Low)
Z.aiLow effort
34.3$0.902/M$0.8568.9 tok/s
#60GPT-5.6 Terra (High)
OpenAIHigh effort
34.2$1.740/M$0.3492.9 tok/s
#61JT-4.1 Flash 236B A21B (Reasoning)
China Mobile
33.9Not listedNot listedNot listed
#62Qwen3.8 27B (Xhigh)
QwenXHigh effort
33.7$0.470/M$1.0141.3 tok/s
#63Motif 3
Motif
33.6Not listedNot listedNot listed
#64Gemini 3.8 Flash (Low)
GoogleLow effort
33.5$0.578/MNot listedNot listed
#65GPT-6 Luna (High)
OpenAIHigh effort
32.9$0.077/M$0.029120.2 tok/s
#66GPT-5.3 Codex (Xhigh)
OpenAIXHigh effort
32.5$1.873/MNot listed100.9 tok/s
#67K2 Horizon 375B A23B
MBZUAI IFM
30.5$0.000/MNot listed891.4 tok/s
#68GPT-5.6 Terra (Medium)
OpenAIMedium effort
30.1$1.740/M$0.1888.4 tok/s
#69Kimi K3 (Low)
KimiLow effort
30.1$2.310/M$1.1542.6 tok/s
#70GPT-6 Luna (Medium)
OpenAIMedium effort
29.9$0.077/M$0.017Not listed
#71Gemini 3.1 Pro Preview
Google
29.7$1.740/M$1.30138.6 tok/s
#72Claude Haiku 5.5 (Low)
AnthropicLow effort
29.5$0.077/M$0.024161.4 tok/s
#73MiniMax-M3
MiniMax
29.2$0.222/M$0.5199.6 tok/s
#74Nex-N2-Pro (Based on Qwen3.5-397B-A17B)
Nex AGI
28.2Not listedNot listedNot listed
#75Solar Pro 4
Upstage
28.2$0.222/MNot listed92.3 tok/s
#76Qwen3.8 27B (Medium)
QwenMedium effort
27.6$0.470/M$1.1344.0 tok/s
#77GPT-5.6 Terra (Low)
OpenAILow effort
27.5$1.740/M$0.1497.8 tok/s
#78JT-4.1 Flash 236B A21B (Non-reasoning)
China Mobile
27.3Not listedNot listedNot listed
#79Quasar 438B (Max, Based on GLM-5.2)
MultiverseMax effort
26.7$0.720/M$2.02134.7 tok/s
#80Apodex 1.1
Apodex
26.4$0.381/M$0.46Not listed

Sources

Updated 2026-10-15. Benchmark methodology and ranking source below. Read how we use the data in themodel ranking methodology.

Artificial Analysis ranking dataArtificial Analysis methodology