Free Vision Models
Free AI APIs that serve vision models - send an image, get text back - ranked by how many image-input models each free tier or credit covers.
Esta sección está en inglés.
A vision model takes an image with your prompt: read a screenshot, describe a photo, pull text out of a scanned page or a chart. This chart ranks the verified free tiers and signup credits that serve at least one model recorded as accepting image input, by how many of those models each one covers; the line under each value names them. Rate limits and model lists differ per provider, so open the offer page before you build on one. To compare vision models one by one, with context size and tool calling, open the model catalog filtered to vision; for free LLM APIs in general, see free LLM APIs.
Los datos de las ofertas se verifican y se muestran en inglés.
| # | Oferta | Vision models | Tarjeta | La letra pequeña |
|---|---|---|---|---|
| 1 | OpenRouter Free AI Models | 9 vision modelsMiniMax M3.1 Flash, Nemotron 3.5 Super VL, Dots3-Note Preview +6 more | No | Free models allow 20 requests/minute and 50/day until you have bought 10 credits (then 1,000/day). A negative balance blocks them, and the lineup rotates (24 variants listed, 17 of them chat models). |
| 2 | Claude API New User Credits | 3 vision modelsClaude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5 | No | Anthropic publishes no amount; it is a one-time small grant that expires (one year after issue unless stated otherwise), then standard paid pricing applies. |
| 3 | OVHcloud AI Endpoints Free Tier | 3 vision modelsQwen3.8-27B, Qwen3.6-27B, Qwen3.5-397B-A17B | Sí | Needs a saved payment method (the US site also requires a $0.99 purchase); billing starts automatically once the credit is used up; anonymous API use is capped at 2 requests per minute. |
| 4 | NVIDIA NIM Free Inference Endpoints | 2 vision modelsnemotron-3.5-super-vl-120b-a12b, deepseek-v4.1-flash | No | Free endpoints are limited to catalog models tagged "Free Endpoint" and can be deprecated; you need a verified NVIDIA account and an API key, and no rate limits or SLA are published for the free tier. |
| 5 | Requesty Free Models | 2 vision modelsgemma-4-31b-it, muse-glimmer-30b | No | Only the 13 designated zero-cost models are free; they are rate-limited per organization and free-model terms can change. |
| 6 | Routeway Free Models | 2 vision modelsGlimmer 30B, Gemma 4 26B MeroMero | No | Free models run on a shared, low-priority queue with best-effort availability and one concurrent request — suitable for prototyping, with no uptime guarantee. |
| 7 | Scaleway Generative APIs Free Tier | 2 vision modelsqwen3.8-27b, qwen3.6-35b-a3b | No | Free Tier applies to your most expensive tokens first; once exceeded, per-token billing starts with a 1,000-token minimum billing unit. |
| 8 | Agnes AI Free API | 1 vision modelagnes-2.5-flash | No | Free video pricing is a limited-time promotion; free access is rate-limited (about 20 effective RPM text, 1 RPM video) and the docs say all limits may be adjusted in the future. |
| 9 | AMD Radeon Cloud Free APIs | 1 vision modelDeepSeek-V4-Flash-Vision-Exp | No | AMD picks which models are served; per-key rate limits, a concurrency allowance, and a spend cap all apply, and only chat-completions and messages endpoints are served. |
| 10 | Cohere Free Trial API | 1 vision modelcommand-a-vision-07-2025 | No | Trial API keys are rate limited and not permitted for production or commercial use. |
| 11 | Infron Free AI Models | 1 vision modeldeepseek-v4.1-flash | No | Standard :free models require a $5+ account balance (not consumed); free requests share one daily pool and run on a shared public pool without SLA. |
| 12 | Jina AI 10M Free API Tokens | 1 vision modeljina-embeddings-v5-omni-small | No | The 10M tokens are one-time and shared across all Jina APIs; a web search costs from 10,000 tokens. The Reader models are CC-BY-NC 4.0, so self-hosting them commercially needs a license. |
| 13 | Pollinations Free API | 1 vision modelopenai-fast | No | Every generation request now requires an API key — there is no documented anonymous tier — and free-tier images may include watermarks. |
| 14 | SmarToken $5 Signup Credit | 1 vision modeldoubao-seed | No | The $5 is promotional balance that cannot be withdrawn as cash, granted once per eligible account, and new Google signups must pass risk checks before 31 Dec 2026 (UTC+8). |
| 15 | Voyage AI Free Embedding and Reranking Tokens | 1 vision modelvoyage-multimodal-3.5 | No | Without a payment method you are held to 3 requests and 10K tokens a minute; the published 2,000 RPM limits need a card on file (Usage Tier 1). Older model generations get no free tokens. |
| 16 | Z.AI Free API Tier | 1 vision modelGLM-4.6V-Flash | No | Free pricing applies only to the Flash models; limited-time free cached-input storage on paid models is promotional and can end. |
| 17 | Cerebras Inference Free Trial | 1 vision modelqwen-3.8-27b | Sí | A verified payment method is required to activate access, and the $5 credits expire 30 days after activation. |
Cómo se ordena esta clasificación
Verified offers of type free tier, fully free, free credits or free trial (startup and student programs are left out because they need an application) whose model list includes at least one model recorded as accepting image input. Sorted by the number of such models, most first; ties go to offers that need no credit card, then by name. Multimodal embedding models that take images count too.
Las filas salen del mismo catálogo que las páginas de ofertas, revisado cada mes. Los valores son los que publica cada proveedor; aifree.dev nunca convierte créditos ni tokens a dólares. Cómo verificamos (en inglés)
Preguntas frecuentes
Esta sección está en inglés.
- What is a vision model?
- A model that accepts an image as part of the input, next to your text prompt, and answers in text: it can describe a photo, read a screenshot or a scanned page, or answer questions about a chart. It is different from an image generator, which takes text and returns a picture.
- Can I send images to a free LLM API?
- Yes, on the providers in this chart, but only to the models recorded as vision models - the line under each value names them. Images usually count as many input tokens, so a free daily limit runs out faster with images than with text.
- Do these free vision APIs need a credit card?
- Most do not. When two offers serve the same number of vision models, the one without a card ranks first, and the Card column shows the requirement for every row.
- How do you know a model accepts images?
- Each offer records, per model, whether the provider documents image input. Models with no recorded answer are not counted, so a provider can serve more vision models than its row shows.