Vision AI models: screenshots, charts and scans as input

Prices from official vendor price lists · October 2026
156 models that read images — screenshots, charts, scanned pages, photos — and reason about what they contain. Output prices run from $0.059 to $180 per 1M tokens across the 137 models with a paid per-token price; 54 have a free tier and 49 ship open weights.
Vision
Services found: 156
#AI modelCapabilitiesAPI $/1M (in / out)Context windowPrice / moOutput $/1MOverviewVisit site
141

Fugu Ultra

Sakana
Text & chatVisionLong context 1M+
$5 / $301M tokensAPI only30View
142

GPT-5.1-Codex-Max

OpenAI
Text & chatVisionCoding
$1.25 / $10400K tokensAPI only10View
143

Seed-2.0-Code

Bytedance Seed
Text & chatVisionVideo
$0.5 / $3262.1K tokensAPI only3View
144

GPT-5 Image Mini

OpenAI
VisionImage generationStructured JSON output
$2.5 / $2400K tokensAPI only2View
145

GPT-5.4 Pro

OpenAI
Text & chatVisionReasoning
$30 / $1801.05M tokensAPI only180View
146

Llama 4 Maverick

Meta
Text & chatVisionLong context 1M+
$0.2 / $0.6961.05M tokensAPI only0.696View
147

Llama 4 Scout

Meta
Text & chatVisionLong context 1M+
$0.1 / $0.310M tokensAPI only0.3View
148

Mistral Large 3

Mistral AI
Text & chatVisionReasoning
$0.5 / $1.5256K tokensAPI only1.5View
149

Qwen3.8 Omni Flash

Qwen
Text & chatVisionAudio
1M tokensAPI onlyView
150

MiMo-V2.6-Pro

Xiaomi
Text & chatVisionAudio
$0.435 / $0.871M tokensAPI only0.87View
151

MiMo-V2.6-Flash

Xiaomi
Text & chatVisionAudio
$0.14 / $0.281M tokensAPI only0.28View
152

MiMo-V2.6-Pro-UltraSpeed

Xiaomi
Text & chatVisionAudio
$4.35 / $8.71M tokensAPI only8.7View
153

Perceptron Mk1.5

Perceptron
Text & chatVisionAudio
$0.15 / $1.536.9K tokensAPI only1.5View
154

Ling 3.0 Flash VL

InclusionAI
Text & chatVisionVideo
$0.02 / $0.059256K tokensAPI only0.059View
155

Pareto

Unbiased
Text & chatVisionPrompt caching
$0.8 / $3.2API only3.2View
156

Ternary Bonsai 2 27B

PrismML
Text & chatVisionReasoning
262.1K tokensAPI onlyView

Browse by need

Reasoning models169 modelsCoding assistants17 modelsVision and images156 modelsLong context 1M+82 modelsFree tier available80 modelsOpen weights88 models

Vision means the model takes images as input and reasons about them - reading a chart, extracting a table from a screenshot, checking a photo against a description. It is not image generation; for that, see the Image generation filter.

156 models qualify, priced from $0.18 to $50 per million output tokens. Image input is billed in tokens too, and a high-resolution photo can consume as much as several pages of text. Claude Fable 5.1, GPT-6 Astra and Claude Opus 5 lead on dense diagrams, handwriting and multi-page scans; for simple tasks the budget tier is a tenth of the price.

Resolution drives costImage input bills in tokens scaled by size. Downscaling before upload is the cheapest optimisation.
Reading is not generatingThese models interpret images; producing them is a separate capability.
Hard cases separate tiersHandwriting, dense diagrams and multi-page scans are where frontier models earn their price.

Models with this capability

The 156 models below all support this capability. Prices are input and output per 1M tokens from the vendor official price list; for dirhams multiply by 3.6725. Tap a card to open the model page.

Claude Fable 5.1
#1
Anthropic
in $10out $501M tokens
GPT-6 Astra
#2
OpenAI
in $10out $501.05M tokens
GPT-6.1 Sol
#3
OpenAI
in $2out $101.05M tokens
GPT-6 Sol
#4
OpenAI
in $2out $101.05M tokens
GPT-6 Luna
#5
OpenAI
in $0.1out $0.51.05M tokens
Claude Opus 5.5
#6
Anthropic
in $4out $201M tokens
Claude Sonnet 5.5
#7
Anthropic
in $2out $101M tokens
Grok 4.7
#8
xAI
in $2out $6500K tokens

How to choose

  • Test on your worst images, not your best — scans and photos of screens are the usual failure point.
  • Check whether the model accepts multiple images per request if you process documents.
  • Compare cost per page, not per token, when the workload is document-heavy.
  • Pair with structured output if you extract fields rather than descriptions.

Common mistakes

Uploading full-resolution photosMost tasks work at a fraction of the pixels and a fraction of the price.
Trusting extracted numbersVision models misread digits under poor contrast. Verify anything financial.

Who this suits

Document processingInvoices, contracts, forms and identity documents at scale.
SupportUsers pasting screenshots instead of describing the problem.
CommerceCataloguing, moderation and attribute extraction from product photos.

Related filters

Image generation9 models that create and edit images
Video understanding53 models that watch and explain video
Frontier tier41 flagship models, the best each vendor sells
Long context 1M+82 models holding 1M tokens or more
All 239 modelsThe full rating with filters and prices

Frequently asked questions

Does vision mean the model draws pictures?

No. Vision is image input. For generating images, see image generation.

How is image input billed?

In tokens, scaled by resolution. A high-resolution photo can cost as much as several pages of text, so downscaling before upload is the first optimisation.

Which model reads handwriting and dense diagrams best?

The frontier tier — Claude Fable 5.1, GPT-6 Astra, Claude Opus 5. For receipts and product photos the budget tier is indistinguishable in practice.

Can vision models read tables?

Yes, and well — but verify totals. Structure survives better than exact figures.

Do they work on video frames?

They accept stills. For continuous footage, use a video-understanding model.

Frequently searched

By capabilityReasoningCodingVisionLong context 1M+Free tierOpen weightsLow latencyRealtime voiceImage generation
By vendorAnthropic (13)OpenAI (30)Google DeepMind (17)xAI (8)Meta (6)DeepSeek (9)Mistral AI (11)Alibaba (Qwen) (34)Z.ai (12)Moonshot AI (5)
Popular modelsClaude Fable 5.1GPT-6 AstraClaude Opus 5Claude Fable 5GPT-5.6 SolGLM 5.3Grok 4.6Kimi K3GPT-5.6 Terra
By deliveryBatch APIPrompt cachingPriority tierFrontier tierProprietary APITool callingStructured JSONWeb searchMultilingual

Ratings 1

Avg. score 5

Rate this page

Help us improve