Short answer (prices checked September 23, 2026): the cheapest usable image APIs cost under $0.02 per image β FLUX.2 [klein] and [dev], GPT Image 2 at low quality, Seedream 5.0 Flash, Imagen 4 Fast and Grok Imagine. Most premium image models cost $0.03β$0.15. For video, 720p prices run from $0.03β$0.05 per second (Veo 3.1 Lite, Runway Gen-4 Turbo) to $0.40 per second for Veo 3.1 with audio.
Where you buy matters as much as which model you pick. FAL is no longer the cheapest route for everything. It halves Replicate's price on Kling v3, but it charges roughly twice BytePlus's direct price for Seedance, and Replicate is cheaper for Seedream 5.0 Pro.
Two shutdowns to plan around: OpenAI's Sora API shuts down on September 24, 2026, and DALL-E 3 already shut down on May 12, 2026.
Skip the integration work. Teamday's image studio runs Seedream 5.0, GPT Image 2.5, Gemini 3.1 Flash Image, Flux 2, Qwen Image and Grok Imagine from one prompt box, and Iris, Teamday's AI designer, can run the same models on a recurring brief. Connect your own provider keys and the prices below are what your provider bills you. Starter is $99/month.
Generate your first image in Teamday βHow to read these prices
- Sources are each provider's own pricing or model page, checked September 23, 2026: OpenAI, Gemini API and Vertex AI, FAL model pages, Replicate model pages, BytePlus ModelArk, Alibaba Model Studio, xAI, and Runway API.
- Image prices are for about one megapixel (1024Γ1024) unless noted. Many providers now bill per megapixel or per token, so larger images cost more.
- Video prices are per second at 720p unless noted. "Audio" means the model generates sound in the same request.
- Seedance is billed in tokens. The per-second figures below are BytePlus's own worked examples (5-second, 16:9 clip) or FAL's stated conversion.
- FAL's general pricing page was out of date on September 23 (it still listed Seedream V4 and Wan 2.5). The prices here come from individual FAL model pages.
Image Generation Pricing
Price per image (about 1 megapixel)
| Model | FAL | Replicate | Direct from model maker | Notes |
|---|---|---|---|---|
| FLUX.2 [klein] 4B | $0.005/MP | $0.001/MP | β | Lowest cost; draft quality |
| FLUX.2 [dev] | $0.012/MP | $0.014/MP | β | Open-weight workhorse |
| FLUX.2 [pro] | $0.03 (first MP) | ~$0.03 ($0.015/run + $0.015/MP) | β | Higher-fidelity Flux |
| FLUX.2 [flex] | $0.05/MP | β | β | Used by Teamday's Flux route |
| GPT Image 2 (low / medium / high) | $0.006 / $0.053 / $0.211 | $0.012 / $0.047 / $0.128 | OpenAI: $0.006 / $0.053 / $0.211 | Strong text rendering |
| GPT Image 2.5 (low / medium / high) | $0.0059 / $0.0132 / $0.0527 | $0.012 / $0.047 / $0.128 | OpenAI: token-billed; $0.0059 low | Newest OpenAI image model |
| Seedream 5.0 Flash | β | β | BytePlus: $0.018 | Cheapest Seedream tier |
| Seedream 5.0 Lite | $0.035 | $0.035 | BytePlus: $0.035 | |
| Seedream 5.0 Pro | $0.0675 | $0.045 | BytePlus: $0.045 | FAL marks its price "tentative" |
| Nano Banana 2 (Gemini 3.1 Flash Image, 1K) | $0.08 | $0.067 | Google: $0.067 | |
| Nano Banana Pro (1K/2K) | $0.15 | $0.15 | Google: $0.134 | |
| Imagen 4 (Fast / Standard / Ultra) | β | β | Vertex AI: $0.02 / $0.04 / $0.06 | Removed from the Gemini API on Aug 17 |
| Grok Imagine Image | β | β | xAI: $0.02 (quality tier $0.05) | |
| Qwen Image 3.0 | $0.04 (1K) | β | Alibaba: $0.03; Pro $0.04 | |
| Wan 2.7 Image | β | β | Alibaba: $0.03; Pro $0.075 | |
| Recraft V4 | $0.04 | $0.04 | β | Vector output costs more |
| Ideogram V4 | $0.0075β$0.025/MP | β | β | Turbo to Quality modes |
Cost of 1,000 images
| Model | Cheapest listed route | Cost per 1,000 |
|---|---|---|
| FLUX.2 [klein] 4B | Replicate | ~$1 |
| GPT Image 2 (low) | OpenAI or FAL | $6 |
| FLUX.2 [dev] | FAL | ~$12 |
| Seedream 5.0 Flash | BytePlus ModelArk | $18 |
| Seedream 5.0 Lite | Any of the three | $35 |
| GPT Image 2 (medium) | Replicate | $47 |
| Nano Banana Pro | $134 |
Takeaway: for images, the direct API is usually cheapest or tied. Replicate undercuts FAL on Seedream 5.0 Pro and GPT Image 2 at medium or high quality; FAL undercuts Replicate on GPT Image 2.5 and the cheapest GPT Image 2 tier.
Video Generation Pricing
Price per second (720p unless noted)
| Model | FAL | Replicate | Direct from model maker | Notes |
|---|---|---|---|---|
| Veo 3.1 Lite | $0.03 no audio / $0.05 audio | $0.05 | Google: $0.05 audio; Vertex $0.03 no audio | Cheapest Veo; 1080p $0.08 with audio |
| Veo 3.1 Fast | $0.10 / $0.15 audio | $0.10 / $0.15 audio | Google: $0.10 audio; Vertex $0.08 no audio | Google direct is cheaper with audio |
| Veo 3.1 | $0.20 / $0.40 audio | $0.20 / $0.40 audio | Google: $0.40 audio; Vertex $0.20 no audio | 4K $0.60 with audio |
| Kling v3 Pro | $0.112 / $0.168 audio | $0.224 / $0.336 audio | β | FAL is half Replicate's price |
| Kling v3 Standard | $0.084 / $0.126 audio | $0.168 / $0.252 audio | β | |
| Seedance 2.5 | ~$0.473 | β | BytePlus: ~$0.231 | Up to 30-second clips |
| Seedance 2.0 | ~$0.30 | $0.18 | BytePlus: ~$0.15 | 1080p ~$0.37 on BytePlus |
| Seedance 2.0 Fast | ~$0.24 | β | BytePlus: ~$0.12 | |
| Wan 3.0 | $0.10 | β | Alibaba: $0.10 (list, 30% off now) | 480p $0.05, 1080p $0.20 |
| Wan 2.6 | β | $0.10 | Alibaba: $0.10 | 1080p $0.15 |
| Grok Imagine Video | β | β | xAI: $0.07 (480p $0.05) | |
| Grok Imagine Video 1.5 | $0.14 | β | xAI: $0.14 | 480p $0.08, 1080p $0.25 |
| MiniMax H3 | $0.06 (768p) | β | β | 480p $0.05 |
| LTX-2.3 | $0.08 (1080p) | $0.08 (1080p) | β | Fast tier $0.06 on FAL |
| Runway Gen-4 Turbo | β | β | Runway: $0.05 | 5 credits/s |
| Runway Gen-4.5 | β | β | Runway: $0.12 | 12 credits/s |
| Runway Aleph 2 | β | β | Runway: $0.28 | Video-to-video editing |
| Sora 2 | $0.10 | $0.10 | OpenAI: $0.10 β shuts down Sep 24, 2026 | Do not start new work |
Cost of 100 videos (5 seconds each, 720p)
| Model | Cheapest listed route | Cost |
|---|---|---|
| Veo 3.1 Lite, with audio | Google, FAL, or Replicate | $25 |
| Runway Gen-4 Turbo | Runway | $25 |
| Wan 3.0 | Alibaba or FAL (list price) | $50 |
| Kling v3 Pro, no audio | FAL | $56 (Replicate: $112) |
| Seedance 2.0 | BytePlus ModelArk | ~$75 (FAL: ~$152) |
| Veo 3.1, with audio | Google, FAL, or Replicate | $200 |
Takeaway: for video, the route decides the bill. Buy Seedance from BytePlus and Kling from FAL; Veo costs the same on Google, FAL and Replicate at standard quality.
Want the clip rather than another integration? Teamday's video studio generates with Seedance 2.5 and 2.0, Kling 3 Pro, Wan and Grok Imagine Video from one prompt, and Reel, Teamday's video producer, can turn a script into a finished cut.
Make your first video in Teamday βPlatform Comparison
Pricing is one axis. For model catalogs, reliability, and API ergonomics, see the AI image & video API providers comparison.
| Platform | Pricing model | Where it is cheapest | Where it is not |
|---|---|---|---|
| FAL | Pay per output (image, megapixel, second, or token) | Kling v3, GPT Image 2.5, broad catalog at first-party prices for Veo and Wan | Seedance (~2Γ BytePlus), Seedream 5.0 Pro, Nano Banana Pro |
| Replicate | Pay per output; some models per second of compute | Seedream 5.0 Pro, FLUX.2 [klein], GPT Image 2 medium/high | Kling v3 (2Γ FAL) |
| OpenAI | Tokens; per-image estimates in its calculator | GPT Image 2 low quality | No video after September 24 |
| Google (Gemini API / Vertex AI) | Per image; per second of video | Nano Banana models, Veo 3.1 Fast with audio | Imagen 4 is Vertex-only now |
| BytePlus ModelArk | Per image; video in tokens | Seedream and Seedance (cheapest route found) | Only ByteDance models |
| Alibaba Model Studio | Per image; per second of video | Wan 3.0 (30% off list at the time of writing), Qwen Image | Free quotas apply in Singapore only |
| xAI | Per image; per second by resolution | Grok Imagine Image at $0.02 | Only Grok Imagine models |
| Runway API | Credits at $0.01 each | Gen-4 Turbo at $0.05/s | Only Runway-hosted models |
Cost Optimization Tips
1. Match the tier to the job
| Use case | Starting model | Cost per image |
|---|---|---|
| Quick mockups | FLUX.2 [klein] or GPT Image 2 low | $0.001β$0.006 |
| Social posts | FLUX.2 [dev] or Seedream 5.0 Flash | $0.012β$0.018 |
| Product photos | Seedream 5.0 Pro or FLUX.2 [pro] | $0.03β$0.045 |
| Text-heavy graphics | GPT Image 2 medium or Qwen Image 3.0 | $0.03β$0.053 |
| Hero images | Nano Banana Pro | $0.134β$0.15 |
2. Check the route before the model
The same model can cost twice as much on a different platform. In this snapshot that applied to Seedance (FAL vs BytePlus) and Kling v3 (Replicate vs FAL).
3. Keep resolution down until the final pass
Per-megapixel and per-resolution billing is now the norm. Nano Banana 2 is $0.067 at 1K and $0.151 at 4K; Seedance 2.0 is about $0.15 per second at 720p and $0.37 at 1080p on BytePlus.
4. Turn audio off when you will replace it
Audio typically adds 50β100%: Kling v3 Pro is $0.112 per second without audio and $0.168 with it on FAL; Veo 3.1 is $0.20 without and $0.40 with.
5. Use batch pricing for bulk work
Google lists Batch at half the standard price for Nano Banana 2 and Nano Banana Pro. Use batch modes for catalogs and variants, not interactive work.
Free Credits
| Platform | What is offered |
|---|---|
| FAL | Free credits and coupons work in the Sandbox and Playground only β not through the API |
| Replicate | Some models run free for a while before billing is required; no amount stated |
| Runware | $2 free credit for business-email signups |
| Alibaba Model Studio | Free image and video quotas in the Singapore region |
Deprecations to Plan Around
| Model | Status | Date |
|---|---|---|
| Sora 2 / Sora 2 Pro and the OpenAI Videos API | Shutdown, no replacement listed | September 24, 2026 |
| DALL-E 2 and DALL-E 3 | Shut down | May 12, 2026 |
| gpt-image-1 | Shutdown | October 23, 2026 |
| gpt-image-1.5, gpt-image-1-mini | Shutdown; replacement gpt-image-2 | December 1, 2026 |
| Nano Banana (gemini-2.5-flash-image) | Shutdown on the Gemini API | October 2, 2026 |
| Imagen 4 | Removed from the Gemini API; still on Vertex AI | August 17, 2026 |
Sources: OpenAI deprecations and the Gemini API changelog.
Recommendation Summary
| Use case | Recommended route |
|---|---|
| Lowest-cost images | FLUX.2 [klein] or [dev] on Replicate or FAL; Seedream 5.0 Flash on BytePlus |
| Text in images | GPT Image 2 or 2.5 (OpenAI or FAL) |
| Cheapest video with audio | Veo 3.1 Lite (Google, FAL, or Replicate) |
| Seedance video | BytePlus ModelArk direct |
| Kling video | FAL |
| Highest-quality Veo | Veo 3.1 β same price on Google, FAL and Replicate |
| Video-to-video editing | Runway Aleph 2 |
You now know the cheapest route for each job. Teamday puts those models in one image studio: pick Seedream, GPT Image, Gemini, Flux, Qwen Image or Grok Imagine, generate, and keep the result in your workspace. Starter is $99/month.
Open the Teamday image studio βRelated Articles
- AI Image & Video API Providers Comparison 2026 β platform-by-platform deep dive
- Best AI Video Models 2026
- Best AI Image Models 2026
- FAL.AI vs Replicate
- AI creative employees β Iris for images and Reel for video
- Build an AI creative team