Kimi K3 added
The newest model in the AIMultiple Intelligence Index is Kimi K3.
LLM Benchmarks
One transparent Intelligence Index combining public benchmarks with AIMultiple's own agentic, RAG and enterprise-reasoning evaluations. See the methodology.
TOP MODEL
Claude Fable 5
Index 92
Best value
MiniMax M3
$0.42 / 1M
Fastest
GPT OSS 120B
0.23s TTFT
Coverage
64 models
8 benchmarks · 336 eval runs
AIMultiple Intelligence Index
Leaderboard
The highest-scoring models across all benchmarks.
Filter & Sort
# | Model | Index | LegalBench | FinanceReasoning | Text-to-SQL | Agentic RAG | ARC-AGI-2 | Agentic LLM Benchmark | FrontierMath | Swe-Bench |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 |  Claude Fable 5 Anthropic | 92 | 89 | 90 | 90 | 98 | - | 69 | - | - |
| 2 |  Kimi K3 Moonshot AI | 88 | 86 | 88 | 77 | 93 | - | 73 | - | - |
| 3 |  GPT-5.6 Sol OpenAI | 87 | 87 | 90 | 74 | 87 | 93 | 62 | - | - |
| 4 |  Grok 4.5 X | 86 | 86 | 88 | 79 | 83 | - | 73 | - | - |
| 5 | GPT-5.5 OpenAI | 83 | 87 | - | - | - | 85 | 59 | - | - |
| 6 | GPT-5.6 Terra OpenAI | 80 | 85 | 87 | 71 | 91 | 84 | 61 | - | - |
| 7 | GPT-5.6 Sol Pro OpenAI | 80 | - | 91 | 79 | 96 | - | 54 | - | - |
| 8 | Claude Opus 4.6 Anthropic | 78 | - | 88 | 68 | 80 | 69 | 72 | - | - |
| 9 |  Gemini 3 Pro Preview Google | 77 | 87 | 86 | 60 | 89 | 31 | - | - | - |
| 10 | Gemini 3.1 Pro Preview Google | 77 | 87 | 87 | 65 | 89 | 77 | 46 | 27 | - |
Page 1 of 7 | ||||||||||
Cost$20.00
Latency4.02s
Context1M
TTFT4.02s
LegalBench
89
FinanceReasoning
90
Text-to-SQL
90
Agentic RAG
98
Agentic LLM Benchmark
69
Cost$6.00
Latency2.93s
Context1M
TTFT2.93s
LegalBench
86
FinanceReasoning
88
Text-to-SQL
77
Agentic RAG
93
Agentic LLM Benchmark
73
Cost$11.25
Latency4.31s
Context1M
TTFT4.31s
LegalBench
87
FinanceReasoning
90
Text-to-SQL
74
Agentic RAG
87
ARC-AGI-2
93
Agentic LLM Benchmark
62
Cost$3.00
Latency7.25s
Context500k
TTFT7.25s
LegalBench
86
FinanceReasoning
88
Text-to-SQL
79
Agentic RAG
83
Agentic LLM Benchmark
73
Cost$11.25
Latency0.70s
Context272k
TTFT0.70s
LegalBench
87
ARC-AGI-2
85
Agentic LLM Benchmark
59
Cost$4.50
Latency1.62s
Context1M
TTFT1.62s
LegalBench
85
FinanceReasoning
87
Text-to-SQL
71
Agentic RAG
91
ARC-AGI-2
84
Agentic LLM Benchmark
61
Cost$11.25
Latency11.11s
Context1M
TTFT11.11s
FinanceReasoning
91
Text-to-SQL
79
Agentic RAG
96
Agentic LLM Benchmark
54
Cost$10.00
Latency1.71s
Context1M
TTFT1.71s
FinanceReasoning
88
Text-to-SQL
68
Agentic RAG
80
ARC-AGI-2
69
Agentic LLM Benchmark
72
Cost$4.50
Latency-
Context1M
TTFT
LegalBench
87
FinanceReasoning
86
Text-to-SQL
60
Agentic RAG
89
ARC-AGI-2
31
Cost$4.50
Latency22.02s
Context1M
TTFT22.02s
LegalBench
87
FinanceReasoning
87
Text-to-SQL
65
Agentic RAG
89
ARC-AGI-2
77
Agentic LLM Benchmark
46
FrontierMath
27
Page 1 of 7
Frontier Over Time
Intelligence Index by model release date
Cost vs Performance
Blended cost against Intelligence Index
Model × Benchmark
Top models on selected benchmarks
Methodology
How The Index is Built
The Intelligence Index normalizes each benchmark to a 0-100 scale and averages a model's relative standing across the benchmarks it was evaluated on. Each benchmark below feeds that score.
AIMultiple
FinanceReasoningComplex multi-step financial reasoning across 238 hard FinanceReasoning questions.AIMultiple
Text-to-SQLNatural-language-to-SQL query generation accuracy across 35+ LLMs.AIMultiple
Agentic RAGMulti-database routing & SQL query generation across 5 databases (BIRD-SQL, 500 questions).AIMultiple
AI MemoryLong-context numeric fact retrieval across document positions (100 items, 14 transcripts).AIMultiple
AI HallucinationResistance to fabricating unmentioned metrics in long-context documents (204 traps, 14 transcripts).Public
Humanity's Last ExamExpert-level, closed-book reasoning across 100+ academic subjects (2.5k questions).Public
SciCodeResearch-level scientific coding across physics, math, biology & chemistry (338 subproblems)Public
LegalBenchCrowd-sourced legal reasoning across 162 tasks from legal practitioners.Public
FrontierMathUnpublished, expert-authored research math problems (Tiers 1–4).Public
ARC-AGI-2Novel visual reasoning puzzles testing generalization, not memorization.Public
Swe-BenchReal-world GitHub issues resolved end-to-end (full 2,294-instance set).Public
LiveCodeBenchContamination-free, continuously updated competitive coding problems.AIMultiple
Agentic LLM BenchmarkAgentic coding across 10 software-dev tasks via CLI tool (~3,500 validation steps).Recent Updates
Recent Updates
Latest changes to the Intelligence Index, model coverage and AIMultiple benchmark methodology.
Moonshot AI
Kimi K3
New model added to the AIMultiple Intelligence Index.
OpenAI
GPT-5.6 Sol
New model added to the AIMultiple Intelligence Index.
OpenAI
GPT-5.6 Terra
New model added to the AIMultiple Intelligence Index.
OpenAI
GPT-5.6 Sol Pro
New model added to the AIMultiple Intelligence Index.
AI Model Release TimelineLast 15 models in 41 days
Key AI model releases from leading providers over the past 41 days.
Provider
21 Jul 2026
Gemini 3.5 Flash Lite +1 moreGemini 3.5 Flash LiteGemini 3.6 Flash
13 Aug 2026
Gemini 3.7 Flash
16 Jul 2026
Muse Spark 1.1
05 Aug 2026
Muse Spark 1.2
09 Aug 2026
Muse Glimmer 30B
24 Jul 2026
Claude Opus 5
16 Jul 2026
Kimi K3
09 Jul 2026
GPT-5.6 Luna +5 moreGPT-5.6 LunaGPT-5.6 Luna ProGPT-5.6 SolGPT-5.6 Sol ProGPT-5.6 TerraGPT-5.6 Terra Pro
08 Jul 2026
Grok 4.5
Explore LLM Use Cases, Analyses & Benchmarks
LLMAug 11
LLM Pricing: Top 15+ Providers Compared
LLM pricing spans three orders of magnitude: the cheapest commodity models cost under $0.20 per million tokens, while frontier reasoning tiers launched as high as $262.50. The chart below tracks how launch prices moved: each model sits at its launch date with its launch list price per million tokens, blended at a 3:1 input-to-output ratio,…
LLM
Benchmark
Aug 11Agentic IT: Can LLMs Design a Benchmark
We gave 12 large language models the job a benchmark team does: invent a benchmark, build it, run four models through it, and report the results. Each did it twice. None of the 24 attempts passed every criterion, and six of the rubric’s checks were passed by none of them. The two topics are text-to-SQL,…
LLM
Benchmark
Aug 9Text-to-SQL: Comparison of LLM Accuracy
We ran 36 large language models over 759 questions from BIRD-SQL, each model writing SQL against a database it had to identify for itself out of 11 candidates. Every parseable query was executed against the real database and its result set compared with the result set of BIRD’s gold query. Missing, malformed and execution-failing queries…
LLM
Open World Evaluation
Aug 6LLM Orchestration in 2026: 22 Frameworks and Gateways
Optimizing LLM orchestration is key to improving performance while keeping resource use under control. To evaluate how different orchestration approaches perform in practice, we benchmarked: Discover selected LLM orchestration tools, including developer frameworks and enterprise gateways: LLM Orchestration involves managing and integrating multiple Large Language Models (LLMs) to perform complex tasks efficiently. It ensures smooth…
LLM
Benchmark
Aug 4HALC-Bench: LLM Hallucination on Long-Context Retrieval Benchmark
HALC-Bench (LLM Hallucination on Long-Context Retrieval Benchmark) measures a large language model’s resistance to fabricating evidence for a metric that does not exist in the target document by using 3 haystacks placed at the beginning, middle, and end of the model’s context window, with 204 questions. claude-fable-5 answered all 204 traps correctly at every haystack…
LLM
Benchmark
Aug 4Compare Multimodal AI Models on Visual Reasoning
We benchmarked 15 leading multimodal AI models on visual reasoning using 200 visual-based questions. The evaluation consisted of two tracks: 100 chart understanding questions testing data visualization interpretation, and 100 visual logic questions assessing pattern recognition and spatial reasoning. Each question was run 5 times to ensure consistent and reliable results. See our benchmark methodology…
LLM
Benchmark
Aug 4Audience Simulation: Can LLMs Predict Human Behavior?
In marketing, evaluating how accurately LLMs predict human behavior is crucial for assessing their effectiveness in anticipating audience needs and recognizing the risks of misalignment, ineffective communication, or unintended influence. Audience simulation with LLMs enables the modeling of virtual audiences, helping organizations anticipate reactions to content or products without relying on costly surveys or focus…
LLM
Benchmark
Aug 2AI Gateways for OpenAI: OpenRouter Alternatives
We benchmarked OpenRouter, SambaNova, TogetherAI, Groq, and AI/ML API across three indicators (first-token latency, total latency, and output-token count), with 300 tests using short prompts (approx. 18 tokens) and long prompts (approx. 203 tokens) for total latency. If you plan to use one of these AI gateways, you can: In this benchmark, we compared OpenRouter,…
LLM
Insight
Jul 16LLM Fine-Tuning Guide for Enterprises
Follow the links for the specific solutions to your LLM output challenges. If your LLM: The widespread adoption of large language models (LLMs) has improved our ability to process human language. However, their generic training often results in suboptimal performance for specific tasks. To overcome this limitation, fine-tuning methods are employed to tailor LLMs to…
LLM
Insight
Jul 12LLM VRAM Calculator for Self-Hosting
Self-hosting an LLM means running inference on hardware the operator controls rather than via a third-party API, which changes the cost, data control, and privacy profile. Whether a model runs at all depends on memory. The calculator estimates the VRAM or unified memory a model needs to run locally, based on the model, its precision,…