Most Context Per Dollar

Models ranked by context window size relative to input token cost. Higher ratios mean more processing capacity for less money.

Methodology: Computed as context_window_tokens / input_price_per_1m. Only models with both values included. Higher is better.

# Model Metric
1 Gemini 1.5 Flash 13,333,333 tokens/$1
2 Gemini 2.0 Flash 6,666,667 tokens/$1
3 Gemini 1.5 Pro 1,600,000 tokens/$1
4 Llama 3.3 70B 1,280,000 tokens/$1
5 GPT-4o mini 853,333 tokens/$1
6 Claude 3 Haiku 800,000 tokens/$1
7 DeepSeek V3 474,074 tokens/$1
8 Qwen 2.5 72B 320,000 tokens/$1
9 Claude 3.5 Haiku 250,000 tokens/$1
10 Mistral 7B 131,072 tokens/$1
11 DeepSeek R1 116,364 tokens/$1
12 Claude 3 Sonnet 66,667 tokens/$1
13 Claude 3.5 Sonnet 66,667 tokens/$1
14 Claude Sonnet 4.6 66,667 tokens/$1
15 Gemini 1.0 Pro 65,536 tokens/$1
16 Grok 2 65,536 tokens/$1
17 GPT-4o 51,200 tokens/$1
18 Mixtral 8x7B 46,811 tokens/$1
19 Mistral Large 2 42,667 tokens/$1
20 Command R+ 42,667 tokens/$1
21 GPT-3.5 Turbo 32,770 tokens/$1
22 Llama 3.1 405B 25,600 tokens/$1
23 o1 13,333 tokens/$1
24 Claude 3 Opus 13,333 tokens/$1
25 GPT-4 Turbo 12,800 tokens/$1
26 Claude 2 12,500 tokens/$1
27 Llama 3 70B 9,102 tokens/$1
28 Llama 2 70B 4,551 tokens/$1
29 GPT-4 273 tokens/$1
30 GPT-3 (davinci-002) 68 tokens/$1
← All leaderboards