AI Models Leaderboard 2026

Compare the latest published AI model snapshot across pricing, context windows, latency, and benchmark performance.

Model	Intelligence	MMLU Pro	Context	GPQA	TTFT (s)	Speed (T/s)	Input Price (1M)	Output Price (1M)
Hermes 4 - Llama-3.1 70B (Non-reasoning) Nous Research Intelligence: 6.7Input (1M): $0.13	6.7	66.4%	-	49.1%	0.58	93	$0.13	$0.40
Mistral: Mistral Small 3 Mistral Intelligence: 6.7Input (1M): $0.10	6.7	65.2%	33K	46.2%	-	-	$0.10	$0.30
Nova Lite Amazon Intelligence: 6.7Input (1M): $0.06	6.7	59.0%	-	43.3%	-	-	$0.06	$0.24
OpenAI: GPT-4o-mini OpenAI Intelligence: 6.7Input (1M): $0.15	6.7	64.8%	128K	42.6%	-	-	$0.15	$0.60
Qwen3 30B A3B (Non-reasoning) Alibaba Intelligence: 6.6Input (1M): $0.20	6.6	71.0%	-	51.5%	-	-	$0.20	$0.80
DeepSeek-V2.5 (Dec '24) DeepSeek Intelligence: 6.5Input (1M): $0.00	6.5	66.6%	-	42.3%	-	-	$0.00	$0.00
Llama 3.1 Instruct 70B Meta Intelligence: 6.5Input (1M): $0.56	6.5	67.6%	-	40.9%	-	-	$0.56	$0.56
Qwen3 4B (Non-reasoning) Alibaba Intelligence: 6.5Input (1M): $0.00	6.5	58.6%	-	39.8%	-	-	$0.00	$0.00
DeepSeek-V2.5 DeepSeek Intelligence: 6.4Input (1M): $0.00	6.4	-	-	-	-	-	$0.00	$0.00
Gemini 2.0 Flash Thinking Experimental (Dec '24) Google Intelligence: 6.4Input (1M): $0.00	6.4	-	-	-	-	-	$0.00	$0.00
Granite 4.1 8BNew IBM Intelligence: 6.4Input (1M): $0.05	6.4	-	-	43.3%	0.47	112	$0.05	$0.10
Sarvam 30B (high)New Sarvam Intelligence: 6.4Input (1M): $0.03	6.4	-	-	63.3%	-	-	$0.03	$0.11
DeepSeek R1 Distill Llama 8B DeepSeek Intelligence: 6.2Input (1M): $0.00	6.2	54.3%	-	30.2%	-	-	$0.00	$0.00
Gemma 4 E2B (Non-reasoning)New Google Intelligence: 6.2Input (1M): $0.00	6.2	-	-	40.5%	-	-	$0.00	$0.00
Mistral: Saba Mistral Intelligence: 6.2Input (1M): $0.00	6.2	61.1%	33K	42.4%	-	-	$0.00	$0.00
Olmo 3.1 32B InstructNew Allen Institute for AI Intelligence: 6.2Input (1M): $0.00	6.2	-	-	53.9%	-	-	$0.00	$0.00
Gemini 1.5 Pro (May '24) Google Intelligence: 6.1Input (1M): $0.00	6.1	65.7%	-	37.1%	-	-	$0.00	$0.00
Olmo 3 32B Think Allen Institute for AI Intelligence: 6.1Input (1M): $0.00	6.1	75.9%	-	61.0%	-	-	$0.00	$0.00
Llama 3.2 Instruct 90B (Vision) Meta Intelligence: 6.0Input (1M): $0.00	6.0	67.1%	-	43.2%	-	-	$0.00	$0.00
Qwen2.5 Turbo Alibaba Intelligence: 6.0Input (1M): $0.05	6.0	63.3%	-	41.0%	-	-	$0.05	$0.20

Showing 501-520 of 793 models