Cheapest LLMs that accept images

Pricing updated daily · Last refreshed Oct 29, 2026

Lowest-priced models that take image input, ranked by blended price per million tokens.

How this list is made

Models whose input types include images, ordered by blended price, lowest first. It ranks by listed price and specifications only. It says nothing about output quality, which this site does not measure, so test candidates on your own task before choosing.

Cheapest LLMs that accept images
ModelProviderBlended $/1MContextInput $/1MOutput $/1M
Llama 4 ScoutMeta$0.15 / 1M1.31M$0.10$0.30
Llama Guard 4 12BMeta$0.18 / 1M164K$0.18$0.18
Claude Haiku 5.5Anthropic$0.20 / 1M1M$0.10$0.50
GPT-6 Luna ProOpenAI$0.20 / 1M1.05M$0.10$0.50
Ministral 3 14B 2512Mistral$0.20 / 1M262K$0.20$0.20
Mistral Small 4Mistral$0.262 / 1M262K$0.15$0.60
Llama 4 MaverickMeta$0.304 / 1M524K$0.188$0.652
Command A+Cohere$0.60 / 1M192K$0.30$1.50
Mistral Large 4Mistral$1.05 / 1M1.05M$0.70$2.10
Gemini 3.6 FlashGoogle$1.50 / 1M1.05M$0.75$3.75

Showing the top 10 of 30 models in the catalogue. Prices are those OpenRouter charges and can differ from a provider’s own list price. Want to sort or filter differently? Use the model comparison tool.

Other rankings

llmtools is a vendor-neutral toolkit for working with large language models — token counting, cost estimation and model comparison, computed in your browser. OpenAI token counts are exact, using the same encoding the model does (o200k_base for GPT-4o, cl100k_base for GPT-4/3.5); Claude and Gemini counts are labelled approximations, since neither publishes a tokenizer that runs client-side. Model pricing is read from OpenRouter’s public catalogue once a day. Nothing you type is ever transmitted anywhere.