--- title: Cheapest AI API 2026 — Prices Compared | ModelsLab description: Cheapest AI API in 2026: compare what each model type costs on one API key, from $0.0047, or run self-hosted models unlimited for $149/mo. url: https://modelslab.com/cheapest-ai-api canonical: https://modelslab.com/cheapest-ai-api type: website component: Seo/CheapestAiApi generated_at: 2026-09-25T12:02:04.380187Z --- AI API pricing Cheapest AI API in 2026 --- One API key for image, video and speech models on our own GPUs. Pay per generation, or pay $149 a month and stop paying per generation on self-hosted models. [Get API key](https://modelslab.com/register) [See pricing](https://modelslab.com/pricing) - Self-hosted models from $0.0047 - Video clips from $0.05 - Speech from $0.001/sec Last updated Sep 25, 2026 · By ModelsLab Engineering AI API prices by modality --- List prices for the models ModelsLab runs on its own GPUs. They are the same on the $21 Basic and $47 Standard plans. Each plan is a dollar allowance, not a call count. ModelsLab self-hosted prices by modality on Basic and Standard, and on the $149 Open Source Unlimited plan | Modality | Self-hosted price | On the $149 plan | |---|---|---| | Image | $0.0047 per image | No per-image charge on self-hosted models | | Video | $0.05 to $0.075 per clip (SVD, CogVideoX, Wan 2.2, LTX-2.3); Wan 2.1 $0.375 per 5 s clip | No per-clip charge on 8 self-hosted models. Partner models (Kling, Veo, Seedance, Sora, Hailuo) bill per second from the wallet | | Speech to text | Whisper large-v3, $0.0047 per request | Unlimited | | Text to speech | $0.001 per second of audio, $0.0047 minimum per request | Unlimited | When a flat plan costs less than paying per call --- The $149 Open Source Unlimited plan has no per-generation charge on self-hosted models. It costs less than list price once your monthly volume passes these points. Monthly volume at which the $149 plan costs less than self-hosted list prices | Workload | List price | $149 costs less above | |---|---|---| | Images | $0.0047 per image | ~31,700 images a month | | Wan 2.2 clips | $0.075 per clip | ~1,990 clips a month | | Speech | $0.001 per second | ~41 hours of audio a month | Above that point, extra self-hosted generations cost nothing more. The limit is 15 parallel generations. Pick the right page for your workload --- ### Image generation Cost per image by monthly volume, and the endpoints to call. - [Cheapest image generation API](https://modelslab.com/cheapest-ai-image-api) - [AI Image Generation API](https://modelslab.com/image-generation-api) ### Video generation Per-clip prices for self-hosted models and per-second prices for partner models. - [Cheapest AI video API](https://modelslab.com/cheapest-ai-video-api) - [Text to Video API](https://modelslab.com/text-to-video-api) ### Speech Text to speech in 48 languages and Whisper large-v3 transcription. - [Cheapest text to speech API](https://modelslab.com/cheapest-text-to-speech-api) - [Speech to Text API](https://modelslab.com/speech-to-text) ### One flat plan Every self-hosted model for $149 a month, 15 parallel generations. - [Unlimited AI API](https://modelslab.com/unlimited) Where ModelsLab is not the cheapest --- LLMs. Our LLMs are partner models, billed per million tokens: from the plan allowance first on Basic and Standard, then the wallet, and from the wallet on the $149 plan. A flat plan does not lower their price. Check the per-token rates before you pick a provider, for example on the [DeepSeek API pricing](https://modelslab.com/deepseek-api-pricing) page. Plans from $21 a month --- Basic is $21, Standard $47 and Open Source Unlimited $149 a month ($210, $451 and $1,500 a year), with 5, 10 and 15 parallel generations. One subscription and one API key cover image, video, speech and LLM. Creating an account is free; API calls need a plan. 100% refund policy on monthly & yearly plans — cancel anytime MonthlyYearly 2 Months Free [Contact Sales](https://calendly.com/support-lael/30min?month=2024-11) Best Value ### Open Source Unlimited Mission-Critical $149 /month Get API Key >>🛡️ 100% refund policy · cancel anytime Unlimited Open Source Models 100% refund policy 24x7 Support 15 parallel generations ⚡ Access to all APIs Unlimited generations on all open-source models For mission critical workloads Add Team Members Priority GPU Clusters Most Popular ### Standard Production $47 /month Get API Key >>🛡️ 100% refund policy · cancel anytime Moderate Traffic 100% refund policy Priority Developer Support 10 concurrent API requests ⚡ For Production workloads API access to all models Prototype ### Basic Prototype $21 /month Get API Key >>🛡️ 100% refund policy · cancel anytime Moderate Traffic 100% refund policy Developer Support via Discord/Email 5 concurrent API requests ⚡ API access to all models Shared GPU Start with the cheapest AI API for your volume --- Pay list price at low volume. Move to $149 when your self-hosted volume passes the break-even point. [Get API key](https://modelslab.com/register) Related Developer APIs --- - [Unlimited AI API](/unlimited)Every self-hosted open-source model for $149/mo. - [AI Image Generation API](/image-generation-api)Text to image, image to image and inpainting on one endpoint. - [Text to Video API](/text-to-video-api)AI video generation from a prompt, self-hosted and partner models. - [Speech to Text API](/speech-to-text)Transcription with word-level timestamps. Get Expert Support in Seconds We're Here to Help. --- Want to know more? You can email us anytime at Chat with support[View Docs](https://docs.modelslab.com) For open-source models, a flat plan is cheaper than per-call pricing once volume grows. On ModelsLab, self-hosted images start at $0.0047 each, self-hosted video clips at $0.05 and text to speech at $0.001 per second of audio. The $149/month Open Source Unlimited plan removes the per-generation charge on every self-hosted model. Prices checked September 2026. $0.0047 per image on self-hosted models, on every plan. At 100,000 images a month on the $149 Open Source Unlimited plan the effective cost is about $0.0015 per image, because that plan has no per-image charge on self-hosted models. On Basic and Standard, self-hosted clips cost $0.05 (SVD, CogVideoX), $0.075 (Wan 2.2, LTX-2.3) or $0.375 for a 5-second Wan 2.1 clip. They have no per-clip charge on the $149 plan. Partner models such as Kling, Veo and Seedance are priced per second. On Basic and Standard they use your plan's included usage first, then your wallet; on the $149 plan they bill from your wallet. Text to speech costs $0.001 per second of generated audio, with a $0.0047 minimum per request. Speech to text (Whisper large-v3) costs $0.0047 per request. Both run on our own GPUs and have no per-request charge on the $149 plan. ElevenLabs and Inworld voices are partner models: on Basic and Standard they use your plan's included usage first, then your wallet. When your self-hosted usage would cost more than $149 a month at list price: about 31,700 images at $0.0047, about 1,990 Wan 2.2 clips at $0.075, or about 41 hours of text to speech at $0.001 per second. Above that, more generations on self-hosted models cost nothing extra; the limit is 15 parallel generations. No. The LLMs on ModelsLab run on partner providers and are billed per million tokens, from your plan's included usage first on Basic and Standard and from your wallet on the $149 plan. The flat-price advantage applies to the open-source image, video and speech models we host on our own GPUs. ModelsLab is a paid service. Plans start at $21/month. Creating an account is free, but API calls require an active plan. ## Frequently Asked Questions ### What is the cheapest AI API in 2026? For open-source models, a flat plan is cheaper than per-call pricing once volume grows. On ModelsLab, self-hosted images start at $0.0047 each, self-hosted video clips at $0.05 and text to speech at $0.001 per second of audio. The $149/month Open Source Unlimited plan removes the per-generation charge on every self-hosted model. Prices checked September 2026. ### How much does one AI image cost? $0.0047 per image on self-hosted models, on every plan. At 100,000 images a month on the $149 Open Source Unlimited plan the effective cost is about $0.0015 per image, because that plan has no per-image charge on self-hosted models. ### How much does an AI video clip cost? On Basic and Standard, self-hosted clips cost $0.05 (SVD, CogVideoX), $0.075 (Wan 2.2, LTX-2.3) or $0.375 for a 5-second Wan 2.1 clip. They have no per-clip charge on the $149 plan. Partner models such as Kling, Veo and Seedance are priced per second. On Basic and Standard they use your plan's included usage first, then your wallet; on the $149 plan they bill from your wallet. ### How much do speech APIs cost? Text to speech costs $0.001 per second of generated audio, with a $0.0047 minimum per request. Speech to text (Whisper large-v3) costs $0.0047 per request. Both run on our own GPUs and have no per-request charge on the $149 plan. ElevenLabs and Inworld voices are partner models: on Basic and Standard they use your plan's included usage first, then your wallet. ### When does the $149 plan cost less than paying per generation? When your self-hosted usage would cost more than $149 a month at list price: about 31,700 images at $0.0047, about 1,990 Wan 2.2 clips at $0.075, or about 41 hours of text to speech at $0.001 per second. Above that, more generations on self-hosted models cost nothing extra; the limit is 15 parallel generations. ### Is ModelsLab also the cheapest LLM API? No. The LLMs on ModelsLab run on partner providers and are billed per million tokens, from your plan's included usage first on Basic and Standard and from your wallet on the $149 plan. The flat-price advantage applies to the open-source image, video and speech models we host on our own GPUs. ### Is there a free AI API? ModelsLab is a paid service. Plans start at $21/month. Creating an account is free, but API calls require an active plan. --- *This markdown version is optimized for AI agents and LLMs.* **Links:** - [Website](https://modelslab.com) - [API Documentation](https://docs.modelslab.com) - [Blog](https://modelslab.com/blog) --- *Generated by ModelsLab - 2026-09-25*