What is the cheapest vision LLM API?

Open-weight multimodal models regularly dip under $0.20 per 1M input tokens. The ranking above updates daily as providers reprice.

How is image input billed?

Images are converted to tokens internally; billing uses the same per-1M-token rates. Actual cost depends on image resolution and the provider's tiling strategy, so treat estimates as approximate.

Can cheap vision models do OCR?

Modern small multimodal models handle printed-text OCR well. Handwriting, complex layouts, and low-quality scans still favor larger models.

Guides · Updated 2026-10-21

Cheapest Vision LLMs (Image Input) — Live Pricing

Vision-capable models read screenshots, PDFs, charts, and photos. These all accept image input, ranked cheapest first — useful for OCR pipelines, UI testing agents, and document processing.

💡 Current leader: Nex AGI: Nex-N2.5-Mini at $0.025 input / $0.100 output per 1M tokens.

Ranking (25 models)

#	Model	Input /1M	Output /1M	Context
1	Nex AGI: Nex-N2.5-Mini nex-agi	$0.025	$0.100	262K
2	OpenAI: GPT-5 Nano (batch) openai	$0.025	$0.200	400K
3	Qwen: Qwen3.7 Flash qwen	$0.030	$0.130	1M
4	ScaleDown: Summarize scaledown	$0.040	Free	1M
5	ScaleDown: Extract scaledown	$0.040	Free	1M
6	Qwen: Qwen3.8 27B qwen	$0.040	$1.35	1M
7	Google: Gemma 3 4B google	$0.040	$0.050	131K
8	DeepSeek: DeepSeek Flash Latest ~deepseek	$0.049	$0.375	1.0M
9	DeepSeek: DeepSeek V4.1 Flash deepseek	$0.049	$0.375	1.0M
10	OpenAI: GPT-5 Nano openai	$0.050	$0.400	400K
11	Anthropic: Claude Haiku 5.5 (batch) anthropic	$0.050	$0.250	1M
12	OpenAI: GPT-6 Luna Pro (batch) openai	$0.050	$0.250	1.1M
13	OpenAI: GPT-6 Luna (batch) openai	$0.050	$0.250	1.1M
14	DeepSeek: DeepSeek V4.1 Flash (batch) deepseek	$0.050	$0.150	1.0M
15	Z.ai: GLM 5.3 Flash (batch) z-ai	$0.050	$0.180	1.0M
16	Google: Gemini 2.5 Flash Lite (batch) google	$0.050	$0.200	1.0M
17	OpenAI: GPT-4.1 Nano (batch) openai	$0.050	$0.200	1.0M
18	Google: Gemma 3 12B google	$0.050	$0.150	131K
19	Z.ai: GLM Flash Latest ~z-ai	$0.052	$0.120	1.0M
20	inclusionAI: Ling 3.0 Flash VL inclusionai	$0.060	$0.180	131K
21	Amazon: Nova Lite 1.0 amazon	$0.060	$0.240	300K
22	Qwen: Qwen3.5-Flash qwen	$0.065	$0.260	1M
23	PrismML: Ternary Bonsai 2 27B prism-ml	$0.075	$0.500	262K
24	Nex AGI: Nex-N2.5-Pro nex-agi	$0.075	$0.250	262K
25	Mistral: Mistral Small 4 (batch) mistralai	$0.075	$0.300	262K

Prices are list prices via OpenRouter, verified 2026-10-21. Provider first-party pricing may differ (volume discounts, batch tiers).

FAQ

What is the cheapest vision LLM API?: Open-weight multimodal models regularly dip under $0.20 per 1M input tokens. The ranking above updates daily as providers reprice.
How is image input billed?: Images are converted to tokens internally; billing uses the same per-1M-token rates. Actual cost depends on image resolution and the provider's tiling strategy, so treat estimates as approximate.
Can cheap vision models do OCR?: Modern small multimodal models handle printed-text OCR well. Handwriting, complex layouts, and low-quality scans still favor larger models.

Estimate your own bill with the cost calculator · Browse all 458 models · More buying guides