IQ of SI
English ▾

Automated page. Written and translated automatically by AI and checked against the sources by software; no human editor. SI-IQ is an index built from public benchmark results, not an IQ test. How it works

How AI got smarter: SI-IQ by release date

Each dot is one AI model, placed by the date it was released and its current SI-IQ. The step line follows the best model available at each point in time.

Closed weights / APIOpen weightsBest so far
SI-IQ of AI models by release date60801001201401602024202520262027100 = reference group averageGPT-6 Astra
Dots: ranked models (blue = closed weights, orange = open weights). Line: best SI-IQ released so far. 100 = average of the 2025 reference group.
Show as a table
ModelDeveloperReleasedSI-IQ
GPT-4o miniOpenAIJul 18, 202468
GPT-4o (Aug 2024)OpenAIAug 6, 202473
GPT-4o (Nov 2024)OpenAINov 20, 202469
Llama 3.3 70BMeta AIDec 6, 202466
o3-miniOpenAIJan 31, 2025105
Claude 3.7 SonnetAnthropicFeb 24, 202599
Gemma 3 27BGoogle DeepMindMar 12, 202577
Llama 4 MaverickMeta AIApr 5, 202581
GPT-4.1OpenAIApr 14, 202591
GPT-4.1 miniOpenAIApr 14, 202591
GPT-4.1 nanoOpenAIApr 14, 202576
o3OpenAIApr 16, 2025116
o4-miniOpenAIApr 16, 2025112
Qwen3-32BAlibabaApr 28, 202595
Claude Opus 4AnthropicMay 22, 202598
Claude Sonnet 4AnthropicMay 22, 202597
Gemini 2.5 Pro (Jun 2025)Google DeepMindJun 5, 2025108
Qwen3-235B-A22B-Thinking (Jul 2025)AlibabaJul 25, 2025111
gpt-oss-120bOpenAIAug 5, 2025106
GPT-5OpenAIAug 7, 2025121
GPT-5 miniOpenAIAug 7, 2025115
GPT-5 nanoOpenAIAug 7, 2025105
Claude Sonnet 4.5AnthropicSep 29, 2025110
Claude Haiku 4.5AnthropicOct 15, 2025104
GPT-5.1OpenAINov 13, 2025120
Claude Opus 4.5AnthropicNov 24, 2025115
GPT-5.2OpenAIDec 11, 2025125
Gemini 3 FlashGoogle DeepMindDec 17, 2025122
GLM-4.7Z.ai (Zhipu AI)Dec 22, 2025112
Kimi K2.5MoonshotJan 27, 2026119
Claude Opus 4.6AnthropicFeb 5, 2026123
Qwen3.5 397B-A17BAlibabaFeb 13, 2026114
Claude Sonnet 4.6AnthropicFeb 17, 2026113
Gemini 3.1 ProGoogle DeepMindFeb 19, 2026127
GPT-5.4OpenAIMar 5, 2026131
GPT-5.4 MiniOpenAIMar 17, 2026117
GPT-5.4 NanoOpenAIMar 17, 2026118
GLM-5.1Z.ai (Zhipu AI)Apr 7, 2026116
Qwen 3.6 35B-A3BAlibabaApr 14, 2026111
Claude Opus 4.7AnthropicApr 16, 2026127
Kimi K2.6MoonshotApr 20, 2026121
Qwen3.6 27BAlibabaApr 22, 2026115
GPT-5.5OpenAIApr 23, 2026138
DeepSeek-V4-ProDeepSeekApr 24, 2026117
Gemini 3.5 FlashGoogle DeepMindMay 19, 2026126
Qwen3.7-MaxAlibabaMay 19, 2026125
Claude Opus 4.8AnthropicMay 28, 2026132
MiniMax-M3MiniMaxJun 1, 2026108
Qwen3.7-PlusAlibabaJun 2, 2026115
Claude Fable 5AnthropicJun 9, 2026138
Kimi K2.7 CodeMoonshotJun 12, 2026118
GLM-5.2Z.ai (Zhipu AI)Jun 16, 2026121
Claude Sonnet 5AnthropicJun 30, 2026126
GPT-5.6 LunaOpenAIJul 9, 2026130
GPT-5.6 SolOpenAIJul 9, 2026140
GPT-5.6 TerraOpenAIJul 9, 2026134
InklingThinking MachinesJul 15, 2026114
Inkling-SmallThinking MachinesJul 15, 2026118
Kimi K3MoonshotJul 16, 2026127
Gemini 3.5 Flash-LiteGoogle DeepMindJul 21, 2026109
Gemini 3.6 FlashGoogle DeepMindJul 21, 2026125
Claude Opus 5AnthropicJul 24, 2026136
DeepSeek V4 Flash 0731DeepSeekJul 31, 2026124
Grok 4.6xAIAug 12, 2026126
DeepSeek V4 Pro 0813DeepSeekAug 13, 2026128
Gemini 3.7 FlashGoogle DeepMindAug 13, 2026131
GLM-5.3Z.ai (Zhipu AI)Aug 14, 2026125
GLM-5.3-FlashZ.ai (Zhipu AI)Aug 20, 2026120
Claude Fable 5.1AnthropicSep 1, 2026140
Gemini 3.8 FlashGoogle DeepMindSep 2, 2026130
GPT-6 AstraOpenAISep 3, 2026150
DeepSeek V4.1 FlashDeepSeekSep 9, 2026128
Claude Opus 5.5AnthropicSep 22, 2026144
GPT-6 LunaOpenAISep 22, 2026130
GPT-6 SolOpenAISep 22, 2026140
Claude Sonnet 5.5AnthropicSep 28, 2026140
GPT-6.1 SolOpenAISep 29, 2026148
Gemini 4 ArgonGoogle DeepMindSep 30, 2026140
Claude Haiku 5.5AnthropicOct 7, 2026129
Muse Spark OpenMeta AIOct 15, 2026127

Record holders

Models that had the highest SI-IQ of any model released up to that date.

ReleasedModelSI-IQGain
Sep 3, 2026GPT-6 AstraOpenAI150+9.6
Jul 9, 2026GPT-5.6 SolOpenAI140+2.1
Apr 23, 2026GPT-5.5OpenAI138+7.3
Mar 5, 2026GPT-5.4OpenAI131+3.7
Feb 19, 2026Gemini 3.1 ProGoogle DeepMind127+1.8
Dec 11, 2025GPT-5.2OpenAI125+4.2
Aug 7, 2025GPT-5OpenAI121+5.0
Apr 16, 2025o3OpenAI116+11.1
Jan 31, 2025o3-miniOpenAI105+32.1
Aug 6, 2024GPT-4o (Aug 2024)OpenAI73+4.4
Jul 18, 2024GPT-4o miniOpenAI68—

By year of release

YearModelsMedian SI-IQBest SI-IQ
202446873
202525106125
202651126150

All scores use the current method version, so older models are re-scored whenever the method changes. Read the full methodology