Best LLM
What is the best LLM for AI agents?
Claude Fable 5.2 leads in agent tasks.
Among the ten best, Claude Sonnet 5.5 is the cheapest at $4 per million tokens.
The fastest among them is Claude Sonnet 5.5, at 103 tokens per second.
- Ranked models
- 100
- Tests
- 12
- Data as of
- Nov 10, 2026
Ranking: AI agents
Place and value in the area, plus overall index, price and speed. Higher values are better.
| Rank | Model | Value | Overall index | Price (3:1) | Speed |
|---|---|---|---|---|---|
| 1 | Claude Fable 5.2Top group | 124.2 | 130.9 | $16 | n/a |
| 2 | Claude Sonnet 5.5Upper group | 120.9 | 121.8 | $4 | 103 |
| 3 | Gemini 4 CarbonUpper group | 119.3 | 127.0 | n/a | n/a |
| 4 | Muse Wonder FullUpper group | 118.6 | 118.1 | n/a | n/a |
| 5 | Claude Opus 5.5Upper group | 116.2 | 125.6 | $8 | 75 |
| 6 | Muse WonderMiddle group | 114.5 | 113.3 | $8 | n/a |
| 7 | Claude Fable 5.1Middle group | 114.0 | 124.1 | $20 | 59 |
| 8 | Gemini 4 ArgonMiddle group | 114.0 | 121.4 | $4 | 46 |
| 9 | Gemini 3.7 FlashMiddle group | 113.4 | 107.9 | n/a | 72 |
| 10 | Claude Opus 5Middle group | 112.5 | 117.4 | $10 | 71 |
| 11 | GPT-6.1 SolMiddle group | 112.3 | 118.0 | $4 | 44 |
| 12 | Kimi K3.1 ProMiddle group | 112.1 | 112.6 | $4 | 56 |
| 13 | GPT-6 AstraMiddle group | 112.1 | 120.2 | $20 | 58 |
| 14 | Claude Fable 5Middle group | 111.4 | 112.8 | $20 | 50 |
| 15 | DeepSeek V4.1 Flash 1017Middle group | 111.0 | 106.7 | $0.26 | 126 |
| 16 | Muse Ember 1.0Middle group | 110.5 | 110.4 | $0.20 | n/a |
| 17 | Grok 4.7Middle group | 108.6 | 103.3 | $3 | 47 |
| 18 | DeepSeek-V4.1-FlashMiddle group | 108.2 | 99.5 | $0.26 | 118 |
| 19 | Claude Haiku 5.5Middle group | 106.9 | 102.8 | $0.20 | 164 |
| 20 | GPT-6 SolMiddle group | 106.7 | 109.1 | $4 | 82 |
Show all 100 ranked models
| Rank | Model | Value | Overall index | Price (3:1) | Speed |
|---|---|---|---|---|---|
| 21 | Ling 3.1 Flash 1019Middle group | 106.2 | n/a | n/a | n/a |
| 22 | GPT-5.6 SolMiddle group | 106.2 | 109.1 | $8 | 75 |
| 23 | MiniMax M3.1 FlashMiddle group | 105.3 | 100.0 | $0.26 | n/a |
| 24 | MiMo V2.6 FlashMiddle group | 105.2 | 97.3 | $0.18 | 38 |
| 25 | Grok 4.6Middle group | 105.0 | 102.1 | $3 | 58 |
| 26 | Claude Opus 4.8Middle group | 104.5 | 99.5 | $10 | 71 |
| 27 | Grok 4.5Middle group | 104.3 | 94.8 | $3 | 52 |
| 28 | MiMo V2.6 ProMiddle group | 104.1 | 100.4 | $0.54 | 32 |
| 29 | GPT-6.1 TerraMiddle group | 103.7 | 106.5 | $2 | 62 |
| 30 | Gemini 3.8 FlashMiddle group | 103.4 | 106.8 | $1.50 | 81 |
| 31 | Muse Spark 1.3Middle group | 102.8 | 107.3 | $2 | 102 |
| 32 | GPT-5.6 TerraMiddle group | 101.9 | 100.8 | $4.50 | 72 |
| 33 | Mistral Large 4Middle group | 101.6 | 100.8 | $1.05 | n/a |
| 34 | GPT-6 LunaMiddle group | 100.5 | 95.8 | $0.20 | 77 |
| 35 | GPT-5.5Middle group | 100.2 | 101.5 | $11.25 | 87 |
| 36 | Step 5 PreviewMiddle group | 100.2 | 97.1 | $1.43 | 65 |
| 37 | GLM-5.3Middle group | 99.9 | 100.6 | $2.15 | 81 |
| 38 | Claude Opus 4.7Middle group | 99.8 | 95.1 | $10 | 57 |
| 39 | Agnes 2.5 Pro ThetaMiddle group | 99.0 | n/a | $0.15 | n/a |
| 40 | GPT-5.6 LunaMiddle group | 98.9 | 93.8 | $0.45 | 123 |
| 41 | Gemini 3.5 FlashMiddle group | 98.4 | 97.8 | n/a | 142 |
| 42 | Qwen 3.8 Max 0902Middle group | 98.3 | 100.8 | $2.48 | 36 |
| 43 | Kimi K3Middle group | 97.4 | 101.1 | $6 | 61 |
| 44 | Muse Spark OpenMiddle group | 96.2 | 97.5 | n/a | n/a |
| 45 | Claude Sonnet 5Middle group | 95.1 | 92.1 | $4 | 77 |
| 46 | DeepSeek-V4-ProMiddle group | 94.4 | 87.2 | $0.99 | 44 |
| 47 | GLM-5.3-FlashMiddle group | 93.3 | 94.3 | $0.24 | 101 |
| 48 | GPT-5.4Middle group | 90.8 | 86.5 | $5.63 | 58 |
| 49 | Gemini 3.1 Pro PreviewMiddle group | 90.4 | 91.4 | $4.50 | 96 |
| 50 | GLM-5.2Middle group | 89.8 | 89.0 | $2.15 | 73 |
| 51 | K2 Horizon 375B-A23BMiddle group | 88.8 | n/a | n/a | n/a |
| 52 | Claude Sonnet 4.6Middle group | 86.8 | 86.7 | $6 | 38 |
| 53 | Qwen 3.8 27BMiddle group | 86.5 | 82.8 | $0.66 | 66 |
| 54 | GLM-5.1Middle group | 85.8 | 82.8 | $2.15 | 39 |
| 55 | Kimi K2.7 CodeMiddle group | 85.5 | 81.3 | $1.71 | 50 |
| 56 | Qwen 3.7 PlusMiddle group | 85.1 | 83.4 | $0.48 | 50 |
| 57 | Reflection BeamLower group | 84.7 | 82.4 | n/a | n/a |
| 58 | Gemini 3 Flash PreviewLower group | 83.6 | 81.0 | $1.13 | 101 |
| 59 | Kimi K2.6Lower group | 83.6 | 84.9 | $1.71 | 32 |
| 60 | Qwen 3.7 MaxLower group | 83.4 | 85.4 | $2.48 | 50 |
| 61 | MiniMax M3Lower group | 80.6 | 83.0 | $0.52 | 82 |
| 62 | GPT-5Lower group | 78.9 | 77.5 | $3.44 | 83 |
| 63 | GPT-5.4 miniLower group | 78.7 | 75.1 | $1.69 | 134 |
| 64 | Qwen 3.5 397B A17BLower group | 75.2 | n/a | $1.29 | 50 |
| 65 | o3Lower group | 75.0 | 74.2 | $3.50 | 73 |
| 66 | GPT-5.4 nanoLower group | 74.4 | 64.0 | $0.46 | 98 |
| 67 | Nemotron 3.5 Super VL 120B A12BLower group | 73.5 | 68.1 | $0.24 | n/a |
| 68 | Grok 4.3Lower group | 70.9 | 71.6 | $1.56 | 66 |
| 69 | Gemini 3.5 Flash-LiteLower group | 70.6 | 73.7 | $0.85 | 145 |
| 70 | Inkling SmallLower group | 70.1 | 68.7 | $0.52 | 61 |
| 71 | Claude Opus 4.5Lower group | 70.0 | 74.4 | $10 | 44 |
| 72 | Claude Sonnet 4.5Lower group | 67.2 | 65.9 | $6 | 38 |
| 73 | MiMo-V2.5-ProLower group | 66.4 | 73.0 | $0.54 | 59 |
| 74 | Kimi K2.5Lower group | 66.3 | 70.3 | $1.14 | 37 |
| 75 | MiniMax M2.7Lower group | 66.1 | 65.1 | $0.37 | 45 |
| 76 | InklingLower group | 64.9 | 72.1 | $1.76 | n/a |
| 77 | o4-miniLower group | 63.1 | 63.5 | n/a | n/a |
| 78 | Muse Glimmer 30BLower group | 61.8 | n/a | $0.52 | 124 |
| 79 | Claude Haiku 4.5Lower group | 60.3 | 54.1 | $2 | 84 |
| 80 | DeepSeek-V3.2Lower group | 60.0 | 48.5 | $0.29 | 31 |
| 81 | Gemini 2.5 ProLower group | 58.6 | 61.4 | $3.44 | 74 |
| 82 | Nemotron 3 UltraLower group | 57.4 | 66.8 | $0.93 | 92 |
| 83 | Mistral Medium 3.5Lower group | 54.4 | 46.3 | $3 | 118 |
| 84 | Ling 3.0 FlashLower group | 54.1 | 47.9 | $0.032 | 106 |
| 85 | Gemini 3.1 Flash-LiteLower group | 52.7 | 55.9 | $0.56 | 125 |
| 86 | GPT-4.1Lower group | 51.6 | 43.6 | $3.50 | 81 |
| 87 | Gemini 2.5 FlashLower group | 47.7 | n/a | $0.85 | n/a |
| 88 | o1Lower group | 44.7 | 49.2 | n/a | n/a |
| 89 | GPT OSS 120BLower group | 39.4 | 43.4 | $0.20 | 92 |
| 90 | Nemotron 3.5 LightningLower group | 37.4 | 31.1 | $0.085 | 180 |
| 91 | Granite 4.2 8BLower group | 37.2 | n/a | n/a | n/a |
| 92 | Mercury 2.5Lower group | 37.0 | 45.3 | n/a | n/a |
| 93 | GLM-4.6Lower group | 36.3 | 45.9 | $0.88 | 28 |
| 94 | Command A PlusLower group | 36.2 | 30.7 | n/a | n/a |
| 95 | North Mini CodeLower group | 32.5 | n/a | n/a | n/a |
| 96 | Granite 4.2 3BLower group | 23.1 | n/a | n/a | n/a |
| 97 | GPT OSS 20BLower group | 21.0 | 30.9 | $0.06 | 85 |
| 98 | Mistral Small 4Lower group | 20.7 | 23.3 | $0.26 | 112 |
| 99 | DeepSeek-V3Lower group | 15.1 | 19.7 | $0.37 | 40 |
| 100 | GPT-4o miniLower group | -1.6 | -7.3 | $0.26 | 66 |
Price in dollars per million tokens, mixed from 3 parts input and 1 part output. Speed in tokens per second.
Claude Fable 5.2Rank 1124.2
Claude Sonnet 5.5Rank 2120.9
Gemini 4 CarbonRank 3119.3
Muse Wonder FullRank 4118.6
Claude Opus 5.5Rank 5116.2
Muse WonderRank 6114.5
Claude Fable 5.1Rank 7114.0
Gemini 4 ArgonRank 8114.0
Gemini 3.7 FlashRank 9113.4
Claude Opus 5Rank 10112.5
GPT-6.1 SolRank 11112.3
Kimi K3.1 ProRank 12112.1
GPT-6 AstraRank 13112.1
Claude Fable 5Rank 14111.4
DeepSeek V4.1 Flash 1017Rank 15111.0
Muse Ember 1.0Rank 16110.5
Grok 4.7Rank 17108.6
DeepSeek-V4.1-FlashRank 18108.2
Claude Haiku 5.5Rank 19106.9
GPT-6 SolRank 20106.7
Which tests count here?
12 tests by independent evaluators feed this area. Each test counts with the result of the model in question.
GDPval-AA
155 models tested · Artificial Analysis
AA-Briefcase
57 models tested · Artificial Analysis
AutomationBench-AA
57 models tested · Artificial Analysis
AA-AnalystAgent
47 models tested · Artificial Analysis
ITBench-AA (SRE)
40 models tested · Artificial Analysis
SkillsBench
32 models tested · Vals AI
APEX-Agents-AA
30 models tested · Artificial Analysis
Agent Poker Bench
9 models tested · Vals AI
GAIA Reliability
9 models tested · Princeton HAL
τ-bench Airline Clean
9 models tested · Princeton HAL
Online Mind2Web
6 models tested · Princeton HAL
AssistantBench
5 models tested · Princeton HAL
Popular comparisons among the best
Other tasks
All LLMs comparedHow the overall index worksCompare several models