Nodaro DocsChoosing a model
Pick the right image, video, audio or text model for each job in Nodaro, from low-cost draft models to premium final renders, with live credit prices.
Nodaro runs more than a hundred image, video, audio and text models. The right one depends on what you make and how much you want to spend. This guide groups the models by job and by price, so you can choose without reading every model page. Start from the quick picks, draft on a low-cost model, and switch to a premium model for the final render. The tables on this page are read from Nodaro's model catalog, so their prices stay current.
Quick picks by task
Find your goal in the left column, then click a model for its settings, prices and prompt tips.
| best for typography / logos / text-heavy | GPT Image 2, GPT Image 2.6 FlareGPT Image 2 for exact copy, labels and non-Latin text; GPT Image 2.6 Flare for posters and layouts, and faster. |
| cheapest realistic image | Z-Image, Qwen, Imagen 4 FastZ-Image is the cheapest. Qwen / Imagen4 Fast for slightly higher quality. |
| highest fidelity image | GPT Image 2, GPT Image 2.6 Sunburst, Imagen 4 UltraGPT Image 2 for characters and identity; GPT Image 2.6 Sunburst for wide or in-between ratios GPT Image 2 can't draw. |
| image edit / restyle | GPT Image 2.6 Flare (I2I), GPT Image 2 (I2I), Seedream 5 Pro (I2I)GPT Image 2.6 Flare changes what you ask and keeps the rest; GPT Image 2 i2i for identity-critical edits; Seedream 5 Pro for instruction-based edits. |
| highest-resolution image | Topaz Image Upscale, GPT Image 2, GPT Image 2.6 SunburstGenerate at the model's top tier, then Topaz upscale 4x (Topaz's only lever is the 1x/2x/4x factor). |
| upscale an image with text, labels or logos | Nodaro Image UpscaleKeeps lettering exactly as written. 1x enhances at the same size; up to 8x. Priced by the output's size, so a small picture costs little. |
| background removal / cutout | Recraft Remove BGCheap, no prompt needed. |
| best cinematic video | VEO 3.1 Quality, Kling 3.0, Seedance 2VEO 3.1 Quality for premium narrative; Kling 3.0 for music-synced motion; Seedance 2 for reference-driven consistency. |
| cheap batch video clips | VEO 3.1 Fast, Wan 2.2 TurboVEO 3.1 Fast is the best price/quality balance with native audio. |
| video with start + end frame | VEO 3.1 Quality, VEO 3.1 Fast, Kling 2.5 Turbo Pro, Hailuo 02 Standard, Seedance 2All listed support an end frame; VEO uses imageUrls[start, end]. |
| music / song generation | Suno V6, Suno V6 Wild, Suno V6 Mini, Suno v5.5V6 is the default flagship; V6 Wild for bolder, less predictable results; V6 Mini when speed matters; v5.5 / v5 / v4 keep their own character. Same price. |
| voice over / narration | ElevenLabs v4, ElevenLabs v3, ElevenLabs v4 Turbo, ElevenLabs Turbo v2.5v4 is the default: [audio tags] for emotion and up to 10,000 characters per request. v3 is the previous expressive model, still selectable. v4 Turbo keeps v4's tags at Turbo's credit price; Turbo v2.5 is the cheapest for plain narration without tags. |
| lip-sync a portrait to audio | Kling Avatar Pro, Kling Avatar StandardPro for best mouth shape; InfiniTalk for resolution control. |
| transcription / captions | ElevenLabs STT, Incredibly Fast Whisper, Whisper, Nemotron 3.5 TranscribeCaptions need WORD timestamps: ElevenLabs STT (always), Incredibly Fast Whisper or Nemotron 3.5 Transcribe (on request, with a speaker per word). Plain Whisper returns phrase segments only. To translate the speech (to or from English) instead of transcribing it, use Nemotron 3.5 Transcribe with a target language; a translation carries no word timings. |
| motion transfer (drive a subject by another video) | Kling 2.6 Motion Transfer, Kling 3.0 Motion TransferKling 2.6 base is cheap; Kling 3.0 is premium. |
Everyday, standard and premium
Models fall into three price tiers. The tiers are relative within one kind of media: each tier holds about a third of the image, video or audio models, split by the price of their default settings. Models with the same price share a tier.
| Tier | Use it for |
|---|---|
| Everyday | Low cost and fast. The right default for drafts, iteration and most work. |
| Standard | A step up in quality for a moderate price. |
| Premium | The best quality at the highest price. Use it for hero shots and final renders. |
Some models have no tier. A model offered under Nodaro's own label, such as Nodaro Image Upscale or Nodaro Video Upscale, is not ranked, and its price does not move any other model's tier. Its description says what its credit price buys.
Compare credits, not tier names, across kinds of media. A premium image model still costs far less than a premium video model.
Prices are for the default settings. The credits in each table below are the price of the model's default settings. A higher resolution, a longer clip or native audio costs more. See Credits for how prices are charged.
Image models
| Job | Model | Why |
|---|---|---|
| Characters, identity and exact text in the image | GPT Image 2 | The default of Generate Image. Best for exact copy, labels and non-Latin text, and for keeping a character's identity. |
| Fast iteration, posters and product shots | GPT Image 2.6 Flare | The fast GPT Image 2.6 model, with strong layouts. Faster than GPT Image 2, and good for posters and layouts with text. |
| Finished work where every detail counts | GPT Image 2.6 Sunburst | The finest detail and the strongest consistency across edits. It also draws the wide and in-between ratios that GPT Image 2 cannot draw, such as 21:9, 4:5 and 5:4. |
| Low-cost drafts and storyboards | Nano Banana 2.1, Z-Image, Seedream 5 Flash | Nano Banana 2.1 is fast and low cost, and takes up to 10 reference images. Z-Image is the cheapest model in the catalog, with a limited set of aspect ratios. Seedream 5 Flash makes 1K or 2K images at the same price. |
| Readable text at a low price | MiniMax H3 Image | Detailed pictures up to about 16 MP at 4K, with readable text inside the picture. 1K and 2K cost the same. |
| Photographic light, skin and dense small text | Seedream 5.5 | More natural light, materials and skin, and dense small text that stays readable. It draws a transparent background in one step. |
| Following detailed instructions | Seedream 5 Pro | The strongest instruction following and visual reasoning of the Seedream family. |
| Editing while keeping the subject | GPT Image 2.6 Flare (I2I), GPT Image 2 (I2I), Seedream 5.5 (I2I) | GPT Image 2.6 Flare changes what you ask and keeps the rest. GPT Image 2 (I2I) suits edits where the identity matters most. Seedream 5.5 (I2I) edits round after round and keeps the rest of the picture as it was. |
| Changing one masked area | GPT Image 2.6 Flare (I2I), GPT Image 2.6 Sunburst (I2I) | Both take an inpainting mask in PNG, JPEG or WebP. The rest of the picture is kept exactly as it was. See Modify Image. |
| Combining many images in one edit | MiniMax H3 Image (I2I) | Changes or combines up to 16 reference images, with readable text. It is priced by size and by each reference image. |
| An animated GIF or a vector SVG | GPT Image 2.6 Flare, GPT Image 2.6 Sunburst | Both write PNG, JPEG, WebP, an animated GIF or an SVG. Choose the file under Format. See Formats in Generate Image. |
| Diagrams and complex compositions | Nano Banana Pro | Strong text rendering, diagrams and complex compositions. |
| The highest resolution | Topaz Image Upscale | Generate at the model's top resolution, then upscale the image 2x or 4x. |
| Upscaling an image with text, labels or logos | Nodaro Image Upscale | Keeps the lettering exactly as written. 1x enhances at the same size, and the factor goes up to 8x. It is priced by the size of the output, so a small picture costs little. |
| Background removal | Recraft Remove BG | Low cost, and no prompt needed. |
If you are unsure between the two GPT Image 2.6 models, draft on Flare and finish on Sunburst. They cost the same.
| Model | Maker | Modes | Credits | Details |
|---|---|---|---|---|
| GPT Image 2 | OpenAI | Text to image | from 21 | Next-gen GPT Image — broader aspect ratios, resolution-based pricing (1K/2K/4K). |
| GPT Image 2.5 Flare | OpenAI | Text to image | from 17 | Fast everyday GPT Image 2.5 - higher quality than GPT Image 2 at about half the latency. The default of the pair: social and creator content, campaign variants, thumbnails, rapid iteration, high-volume work. |
| GPT Image 2.5 Sunburst | OpenAI | Text to image | from 17 | Precision GPT Image 2.5 - trades generation time for tighter control and detail fidelity. Pick it for brand-sensitive and production work: packaging, diagrams, ecommerce retouching, polished campaign creative. |
| Nano Banana 2.1 | Text to image, Image to image | from 11 | Google's high-efficiency image model for generation and editing — sharper quality, prompt adherence, character consistency and text rendering, up to 10 references and 4K output. | |
| Z-Image | Tongyi-MAI | Text to image | 3 | Cheapest model in catalog. Fast, stylized output. Limited aspect ratios. |
| Seedream 5 Flash | Bytedance | Text to image | 11 | Fast, low-cost Seedream 5 — quick generations at 1K or 2K, same price at either size. |
| Seedream 5 Pro | Bytedance | Text to image | from 20 | Flagship Seedream 5 Pro — strongest instruction following and visual reasoning. Basic = 1K, high = 2K. |
| GPT Image 2.5 Flare (I2I) | OpenAI | Image to image | from 17 | Fast GPT Image 2.5 edits (up to 16 source images) - the default when you are iterating rather than finishing. |
| GPT Image 2 (I2I) | OpenAI | Image to image | from 21 | Image-to-image with GPT Image 2. |
| Nano Banana Pro | Text to image, Image to image | from 50 | Top-tier Nano Banana — best for text rendering, diagrams, and complex compositions. | |
| Topaz Image Upscale | Topaz | Image upscaling | from 28 | High-quality image upscale at 1x (enhance only), 2x or 4x. Best for production-ready output. |
| Nodaro Image Upscale | — | Image upscaling | from 5 | Image upscale at 1x (enhance only), 2x, 4x or 8x. Restores detail and keeps lettering exactly as written; the output is the input times the factor, nothing cropped. Priced by the output's size. |
| Recraft Remove BG | Recraft | Background removal | 4 | Remove image background. Cheap utility. |
Every image model is on Image models.
Video models
Before you choose, check three things on the model page. Does the model make native audio? Which durations does it offer? Does it accept an end frame as well as a start frame? Many video models make both text-to-video and image-to-video under one name. Generate Video chooses the mode from what you connect: with a start frame it animates the image, without one it works from the prompt.
| Job | Model | Why |
|---|---|---|
| Most clips | VEO 3.1 Fast | The best balance of price and quality, with native audio, in 4, 6 or 8 seconds. |
| Many clips at the lowest price | VEO 3.1 Lite, Wan 2.2 Turbo | Built for high volume, such as B-roll. |
| Premium, cinematic shots | VEO 3.1 Quality | Premium narrative video with native audio and an optional end frame. |
| Motion synced to music | Kling 3.0 | Premium motion, 3 to 15 seconds, with native audio. |
| A faster, lower-cost Kling | Kling 4.0 Flash | The faster Kling 4.0 at 720p, up to 20 seconds, with native audio. It works from a start frame, reference images, videos and characters, or plain text. |
| Consistency from reference images | Seedance 2 | Reference-driven video with native audio. Seedance 2 Mini is the budget tier. |
| One long shot | Seedance 2.5, Wan 3.0 | Up to 30 seconds in one shot. |
| A start frame and an end frame | VEO 3.1 Quality, VEO 3.1 Fast, Kling 2.5 Turbo Pro, Seedance 2 | Each supports an end frame. |
| A talking portrait | Kling Avatar Pro, Kling Avatar Standard | Kling Avatar Pro gives the best mouth shapes. |
| Motion from another video | Kling 2.6 Motion Transfer, Kling 3.0 Motion Transfer | Kling 2.6 is the low-cost option. Kling 3.0 is premium. |
| A longer clip from any video | Seedance 2.5 Extend, Seedance 2.0 Extend, Kling 4 Extend | Each makes a continuation with audio and stitches it into one clip. Seedance 2.5 Extend adds 4 to 30 seconds, Seedance 2.0 Extend adds 4 to 15 seconds, and Kling 4 Extend adds 3 to 30 seconds. See Extend Video. |
| A sharper final video | Topaz Video Upscale, Nodaro Video Upscale | Topaz gives a high-quality upscale and enhancement to 720p, 1080p or 4K, at up to 60 fps. Nodaro Video Upscale takes any clip of up to 60 seconds and 50 MB to 1080p, 2K or 4K, and keeps its shape, frame rate, length and sound. It is priced per second of the clip. |
| Model | Maker | Modes | Credits | Details |
|---|---|---|---|---|
| VEO 3.1 Lite | Image to video, Text to video | from 83 | VEO 3.1 Lite — most cost-effective VEO tier for high-volume generation. 4/6/8s with audio, supports first+last frame. | |
| VEO 3.1 Fast | Image to video, Text to video | from 165 | VEO 3.1 Fast — cheaper VEO 3.1 tier, 4/6/8s with audio. Good balance for most uses. Flat per-generation pricing across durations. | |
| VEO 3.1 Quality | Image to video, Text to video | from 1,100 | Google VEO 3.1 Quality — premium cinematic video. 4/6/8s clips, optional end frame, native audio. No reference-to-video mode (Fast/Lite only). Flat per-generation pricing across durations. | |
| Kling 3.0 | Kuaishou | Image to video, Text to video | from 297 | Premium Kling 3.0 — variable 3-15s duration, native audio, 720P/1080P. |
| Seedance 2 Mini | Bytedance | Image to video, Text to video | from 132 | Budget Seedance 2 tier — 480p/720p only, per-second pricing by resolution. |
| Seedance 2 | Bytedance | Image to video, Text to video | from 253 | Seedance 2 — premium tier with native audio. Per-second pricing by resolution. |
| Seedance 2.5 | Bytedance | Image to video, Text to video, Video to video | from 374 | Seedance 2.5 — up to 30s in one shot, native audio, wide multimodal references. 480p/720p/1080p. |
| Wan 2.2 Turbo | Alibaba | Image to video, Text to video | from 110 | Cheap, fast Wan turbo — 5s. Serves both i2v and t2v under one id. |
| Kling 2.5 Turbo Pro | Kuaishou | Image to video, Text to video | from 121 | Faster Kling — good quality at lower cost. Supports end frame. |
Every video model is on Video models. For how two of them, Seedance 2.5 and Seedance 2 Fast, follow camera directions, read Camera Motion Lab.
Audio, voice and music models
| Job | Model | Why |
|---|---|---|
| Voice-over with emotion | ElevenLabs v4, ElevenLabs v3 | ElevenLabs v4 is the default. It reads audio tags in brackets for emotion and pacing, and takes up to 10,000 characters per request. ElevenLabs v3 is the previous expressive model, and you can still choose it. |
| Narration at a lower price | ElevenLabs v4 Turbo, ElevenLabs Turbo v2.5 | ElevenLabs v4 Turbo keeps the audio tags of v4 at the credit price of Turbo v2.5. Turbo v2.5 is the cheapest for plain narration without tags. |
| Speech in many languages | ElevenLabs Multilingual v2 | Built for multiple languages. |
| A conversation between several voices | ElevenLabs Dialogue v3 | Voices each role of a script. |
| Sound effects | ElevenLabs Sound Effects | Short effects from a text prompt, at a very low price. |
| Songs and music | Suno V6 | The default. Suno V6 Wild is bolder and less predictable, Suno V6 Mini is faster. All Suno models cost the same. |
| Transcripts for captions | ElevenLabs STT, Incredibly Fast Whisper, Nemotron 3.5 Transcribe | Captions need a time for every word, and these return it. Nemotron 3.5 Transcribe returns word times on request, with a speaker for each word. Whisper returns phrase timing only. |
| Translating speech | Nemotron 3.5 Transcribe | Translates the speech instead of transcribing it, to or from English. A translation has no word times. It is priced per started minute of the file, up to 60 minutes. See Transcribe. |
| Model | Maker | Modes | Credits | Details |
|---|---|---|---|---|
| ElevenLabs v4 | ElevenLabs | Text to speech | from 5 | Newest ElevenLabs TTS — [audio tags], stability and similarity control, up to 10,000 characters per request. Direct API. |
| ElevenLabs v3 | ElevenLabs | Text to speech | from 5 | Expressive ElevenLabs TTS — supports [audio tags] for emotion / pacing. Direct API. |
| ElevenLabs v4 Turbo | ElevenLabs | Text to speech | from 3 | Faster, cheaper ElevenLabs v4 at lower fidelity — [audio tags], stability and similarity control, at Turbo's credit price. Prefer v4 when quality matters. Direct API. |
| ElevenLabs Turbo v2.5 | ElevenLabs | Text to speech | from 3 | Fast, cheap ElevenLabs TTS via the direct ElevenLabs API. Good for narration. |
| ElevenLabs Multilingual v2 | ElevenLabs | Text to speech | from 5 | Multi-language ElevenLabs TTS via the direct ElevenLabs API. |
| ElevenLabs Dialogue v3 | ElevenLabs | Multi-speaker dialogue | from 5 | Multi-speaker dialogue via the direct ElevenLabs API — give it a script, it voices each role (any voice: premade, library, or cloned). |
| ElevenLabs Sound Effects | ElevenLabs | Sound effects | from 2 | Generate short sound effects from a text prompt. |
| ElevenLabs STT | ElevenLabs | Speech to text | 25 | Speech-to-text with WORD-level timestamps (always on), speaker diarization and audio-event tags. The engine to use when the transcript feeds captions. |
| Incredibly Fast Whisper | OpenAI | Speech to text | 44 | Fast Whisper speech-to-text. Returns WORD-level timestamps when asked, so its transcript can feed captions. |
| Whisper | OpenAI | Speech to text | 44 | Whisper speech-to-text — PHRASE-level segments only, NO word timestamps. Fine for a transcript or a static subtitle; not for word-timed (kinetic) captions. |
| Suno V6 | Suno | Music | 33 | Suno V6 — greater musical expression with more natural vocals and richer details. The flagship and the default. |
| Suno V6 Wild | Suno | Music | 33 | Suno V6 Wild — pushes creative boundaries for bolder, more distinctive musical expression; more varied, less predictable results. |
| Suno V6 Mini | Suno | Music | 33 | Suno V6 Mini — lightweight and fast, balancing quality and speed for effortless creation. |
Every audio model is on Audio models.
Text models
Nodes that write or read text with a language model let you choose the model. Examples are Generate Text - Any LLM, Generate Script, QA Check, Describe Image, Describe to Picker, Image Critic and the AI judge of Choose Best. The model selector groups the models into three tiers, as a guide to quality and budget:
| Tier | Use it for | Relative price per token |
|---|---|---|
| Economy | High-volume, simple text: fast drafts, short prompts, checks. | Lowest |
| Standard | Most writing and structured output. The balance of quality and price. | Middle |
| Premium | The hardest reasoning: long scripts, detailed scene plans, careful rewrites. | Highest |
Each call is charged for the tokens it actually used, at the rate of the model that ran it. The Run button shows ≈ and a number: an estimate of a typical call on that model. A long prompt or a long answer costs more on any model, and a higher Reasoning Effort costs more only through the extra tokens it writes. On GPT-6.1 Terra, a typical run has the same estimate at every Reasoning Effort. The full rules, including the hold before a call and what a failed call costs, are in How LLM calls are charged.
The models on offer differ by node. Each node names its default model on its own page.
Draft cheap, finish premium
Draft on an everyday model
Build the workflow on a low-cost model at a low resolution or a short duration. Run it until the composition, the prompt and the pickers are right.
Keep the model in one node
Drive similar shots from a List into one generation node instead of copying the node for every shot. The node runs once per list item, so moving from the draft model to the final one is a single change.
You can also connect one Provider node to the Settings input of every Generate Image node, or of every Generate Video or Generate Video Pro node. The Provider node then sets the model of all of them.
Render the final version
Switch to a premium model or a higher resolution, and run again. Compare the result with the draft before you spend on the rest of the workflow.
Tips
- Use the right model for text in images. For a sign, a label or a poster with words, use GPT Image 2 or GPT Image 2.6 Flare.
- Check the video features first. A model without native audio needs a separate voice or music track. A model without end-frame support cannot land on a planned last frame.
- Save money on retries. Drafts on an everyday model cost a fraction of a premium run, so do your trial and error there.
Frequently asked questions
Which image model should I start with?
Start with GPT Image 2, the default of Generate Image, for characters, identity and exact text in the image. Use GPT Image 2.6 Flare to iterate fast and for posters and product shots, and GPT Image 2.6 Sunburst when every detail counts. Use Nano Banana 2.1 or Z-Image for low-cost drafts.
Which video model gives the best results?
VEO 3.1 Quality for premium, cinematic shots, Kling 3.0 for music-synced motion, and Seedance 2 for consistency driven by reference images. VEO 3.1 Fast is the best balance of price and quality for most clips, with native audio.
Which model should I use to make captions?
Use ElevenLabs STT, Incredibly Fast Whisper or Nemotron 3.5 Transcribe, because captions need a time for every word. Nemotron 3.5 Transcribe also gives each word a speaker. Plain Whisper returns only phrase-level timing, which is fine for a transcript but not for word-by-word captions.
Why is a model's price shown as a starting price?
The number in the tables is the price of the model's default settings. A higher resolution, a longer clip or native audio costs more. Every model page lists all of its prices.
Can I switch the model of many nodes at once?
For Generate Image and the video nodes, yes. Connect one Provider node to the Settings input of each Generate Image node, or of each Generate Video or Generate Video Pro node, then change the model in the Provider node. For other nodes, drive similar shots from a List into one generation node, which runs once per item with the same model.
Related
AI models in Nodaro
Credits
Generate Image
Generate Video
Lists and batching
Last updated on
Character training
Train a custom model on a character's pictures in Character Studio for the closest likeness in Generate Image, for 1,650 credits in about 15 minutes.
Reference boards and consistency grids
Build a reference board or a consistency grid from one photo, pick the right model, and reuse it so a person, product or place looks the same everywhere.