AI Battle > Data

AI Battle Data

Open JSON for every ranking and model. The same numbers that drive the human pages, in a flat format agents, search engines and reporters can pull directly. Refreshed daily with the rest of the site.

Rankings (129 files)

• End-to-End Response Time — LLM Models
• AA-Briefcase Elo — LLM Models
• AA-Omniscience Index — LLM Models
• Output Tokens per Intelligence Index Task — LLM Models
• Cost per Intelligence Index Task — LLM Models
• Cost to Run Artificial Analysis Intelligence Index — LLM Models
• Pricing: Cache Hit, Input, and Output — LLM Models
• Context Window — LLM Models
• Output Speed — LLM Models
• Time per Intelligence Index Task — LLM Models
• Latency: Time To First Answer Token — LLM Models
• Model Size: Total and Active Parameters — LLM Models
• Artificial Analysis Openness Index: Score — LLM Models
• Artificial Analysis Intelligence Index by Open Weights / Proprietary — LLM Models
• Speed — LLM Models
• Cost per Task — LLM Models
• Intelligence — LLM Models
• GDPval-AA v2 Leaderboard — LLM Index
• Artificial Analysis Intelligence Index by Open Weights / Proprietary — LLM Index
• Artificial Analysis Intelligence Index — LLM Index
• Artificial Analysis Finance & Accounting Index — LLM Index
• Cost per Intelligence Index Task — LLM Index
• AA-Omniscience Index — LLM Index
• AA-Briefcase Elo — LLM Index
• Cost to Run Artificial Analysis Intelligence Index — LLM Index
• Pricing: Cache Hit, Input, and Output — LLM Index
• Output Speed — LLM Index
• Time per Intelligence Index Task — LLM Index
• Output Tokens per Intelligence Index Task — LLM Index
• Output Speed: gpt-oss-120b (high) — LLM Index
• Artificial Analysis Intelligence Index — LLM Models
• Pricing (Cache Hit, Input, and Output): gpt-oss-120b (high) — LLM Index
• Endpoint Accuracy Index: gpt-oss-120b (high) — LLM Index
• Provider Voice Arena Quality Elo — LLM Index
• Text to Image Leaderboard — LLM Index
• Artificial Analysis Openness Index: Components — LLM Index
• Speed — LLM Index
• AA-AnalystAgent pass^5 — LLM Index
• Cost per Task — LLM Index
• Intelligence — LLM Index
• API Generation Time — Image Generation
• API Generation Time, Variance — Image Generation
• Artificial Analysis Image Arena Quality Elo — Image Generation
• Price ($/1k images) — Image Generation
• Image Arena Quality Elo — Image Generation
• API Price — Image Generation
• Characters Per Second — Text-to-Speech
• Price — Text-to-Speech
• Price of Transcription — Speech-to-Text
• Provider Voice Arena Quality Elo — Text-to-Speech
• Speed Factor — Speech-to-Text
• AA-WER (Non-streaming): AA-AgentTalk Dataset — Speech-to-Text
• Artificial Analysis Word Error Rate Index (Non-streaming) — Speech-to-Text
• VoxPopuli: Cleaned vs Original Subset of Publicly Available Data — Speech-to-Text
• WER Index (Non-streaming) — Speech-to-Text
• Price — Speech-to-Text
• WER Index - Final Transcription — Streaming Speech-to-Text
• Latency — Streaming Speech-to-Text
• Price — Streaming Speech-to-Text
• Arena Elo — Video Editing
• Arena Elo — Video Generation
• Arena Elo — Image-to-Video
• Arena Elo — Instrumental Music
• Arena Elo — Music Vocals
• Arena Preference Elo — Speech-to-Speech
• Time to First Audio — Speech-to-Speech
• Conversational Dynamics (Full Duplex Bench subset) - Category Breakdown — Speech-to-Speech
• Conversational Dynamics (Full Duplex Bench subset) — Speech-to-Speech
• Speech Reasoning (Big Bench Audio) — Speech-to-Speech
• Cost per Hour of Input Audio (Big Bench Audio subset) — Speech-to-Speech
• Agentic Performance (𝜏-Voice) by Domain — Speech-to-Speech
• Artificial Analysis Speech to Speech Index — Speech-to-Speech
• Speed — Speech-to-Speech
• Agentic Performance (𝜏-Voice) — Speech-to-Speech
• Cost per Hour of Input Audio — Speech-to-Speech
• AA-Speech to Speech Index — Speech-to-Speech
• Arena Elo — Image Editing
• Arena Elo — Text-to-Image
• Coding Agent Index — Coding Agents
• Cost per Task — Coding Agents
• Time per Task — Coding Agents
• Artificial Analysis Intelligence Index: Max intelligence supported model — AI Chatbots
• Monthly Price: Paid Chatbot Plans — AI Chatbots
• Price of Paid Plans — AI Chatbots
• Feature score of Paid Plans — AI Chatbots
• Intelligence of Paid Plans — AI Chatbots
• Humanity's Last Exam: Cost per Task — Frontier Evals
• Humanity's Last Exam: Output Tokens per Task — Frontier Evals
• Humanity's Last Exam: Score — Frontier Evals
• Humanity's Last Exam: Time per Task — Frontier Evals
• MMMU-Pro: Total Cost to Run — eval mmmu
• MMMU-Pro: Token Usage — eval mmmu
• Terminal-Bench v4.0: Cost per Task — eval terminal
• MMMU-Pro: Score — eval mmmu
• Terminal-Bench v4.0: Time per Task — eval terminal
• Terminal-Bench v4.0: Score — eval terminal
• Terminal-Bench v4.0: Output Tokens per Task — eval terminal
• AA-Omniscience Hallucination Rate — eval omniscience
• AA-Omniscience Index: Cost Breakdown — eval omniscience
• AA-Omniscience Index Across Domains (Normalized) — eval omniscience
• AA-Omniscience Index: Token Usage — eval omniscience
• Software Engineering AA-Omniscience Index Across Languages (Normalized) — eval omniscience
• AA-Omniscience Index: Score — eval omniscience
• AA-Omniscience Accuracy — eval omniscience
• AA-Briefcase Tool Calls Breakdown, Avg per Task — eval briefcase
• AA-Briefcase Rubric Pass Rate by File Type (Normalized) — eval briefcase
• AA-Briefcase Elo — eval briefcase
• AA-Briefcase Cost per Task — eval briefcase
• AA-Briefcase Output Tokens per Task — eval briefcase
• Time per Task — eval briefcase
• Mean Turns per Task — eval briefcase
• AA-LCR v1.1: Output Tokens per Task — eval lcr
• AA-LCR v1.1: Score — eval lcr
• AA-LCR v1.1: Time per Task — eval lcr
• AA-LCR v1.1: Cost per Task — eval lcr
• Artificial Analysis Openness Index: Components — eval openness
• Artificial Analysis Openness Index: Score — eval openness
• GDPval-AA v2 Leaderboard — eval gdpval
• GDPval-AA v2: Output Tokens per Task — eval gdpval
• GDPval-AA v2: Average Turns per Task — eval gdpval
• GDPval-AA v2: Cost per Task — eval gdpval
• 𝜏³-Banking: Time per Task — eval tau3
• 𝜏³-Banking: Score — eval tau3
• 𝜏³-Banking: Cost per Task — eval tau3
• 𝜏³-Banking: Output Tokens per Task — eval tau3
• SciCode: Time per Task — eval scicode
• SciCode: Output Tokens per Task — eval scicode
• SciCode: Cost per Task — eval scicode
• SciCode: Score — eval scicode

Model profiles (65 files)

• Gemini 3.5 Flash-Lite
• gpt-oss-120b (high)
• DeepSeek V4.1 Flash (Reasoning, Max Effort)
• Nemotron 3 Ultra 550B A55B (Reasoning)
• Gemini 3.8 Flash (high)
• Mistral Medium 3.5
• MiniMax-M3
• GLM-5.3-Flash
• Muse Glimmer (high)
• DeepSeek V4 Pro 0813 (Reasoning, Max Effort)
• Inkling (xhigh)
• Muse Spark 1.3 (max)
• GLM-5.3 (max)
• Grok 4.6 (high)
• Claude Opus 5 (Adaptive Reasoning, Max Effort)
• Qwen3.8 27B (xhigh)
• Qwen3.8 2.4T A95B
• Kimi K3 (max)
• Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)
• GPT-5.6 Sol (max)
• Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)
• GPT-6 Astra (max)
• GPT-5.6 Luna (max)
• GPT-5.6 Terra (max)
• K2 Horizon 375B A23B
• Claude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)
• GPT-6 Astra (xhigh)
• Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)
• Claude Opus 5 (Adaptive Reasoning, Xhigh Effort)
• Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)
• Gemini 3.7 Flash (low)
• Gemini 3.7 Flash (high)
• GPT-6 Astra (medium)
• GPT-6 Astra (high)
• Qwen3.8-Flash-Next
• Claude Opus 5 (Adaptive Reasoning, High Effort)
• GPT-5.5 (high)
• GPT-5.5 (xhigh)
• Gemini 3.8 Flash (medium)
• K2 Think V2
• K2-V2 (high)
• K2-V2 (medium)
• K2-V2 (low)
• Olmo 3.1 32B Think
• Olmo 3.1 32B Instruct
• Olmo 3 7B Think
• Molmo 7B-D
• Olmo 3 7B Instruct
• Apertus 70B Instruct
• Apertus 8B Instruct
• Nemotron 3.5 Lightning
• Nemotron 3 Super 120B A12B (Reasoning)
• Nemotron Cascade 2 30B A3B
• NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)
• MiniCPM5-1B (Reasoning)
• MiniCPM5-1B (Non-reasoning)
• NVIDIA Nemotron 3 Nano 4B
• NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)
• Grok 4.6 (xhigh)
• Muse Spark 1.3 (xhigh)
• Agnes 3.0 Flash
• Qwen3.8 Max
• Qwen3.8 27B (medium)
• Gemini 3.7 Flash (medium)
• Muse Spark 1.1 (xhigh)

Citation: AI Battle, "AI Battle data" + dataset name, 2026-09-19.

Method

Each file is rebuilt every deploy; values are not post-processed beyond the statistics shown on the page they mirror — see the methodology for sources, limitations and corrections.