Tutti i modelli e le utilità multimediali disponibili su Gobostock.

Flux Dev is an image generation model available in Gobostock. High quality, balanced speed. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 7 credits, and the final quote is shown before anything is charged.

P-Image is Pruna AI’s real-time text-to-image model, tuned for low-latency generation, prompt adherence, and reliable text rendering. Gobostock uses it as an economical choice for fast production drafts and photorealistic concepts when creators need many iterations before selecting a final stock or campaign direction.

Z-Image-Turbo is a distilled Alibaba vision model designed for very fast image generation while retaining sharp photorealistic output, complex layout following, and Chinese and English text rendering. Gobostock exposes its seed, steps, scheduler, and optional quality-boost controls for economical iterative creation.

FLUX.2 Dev is an open-weight Black Forest Labs image model for text-to-image and reference-guided image generation. Gobostock uses it when creators want diffusion-style controls such as steps, guidance, scheduler choice, seed reuse, negative prompts, LoRAs, and up to four reference images.

Flux Schnell is an image generation model available in Gobostock. Fastest generation. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 4 credits, and the final quote is shown before anything is charged.

Seedream 4.0 is an image generation model from ByteDance, available in Gobostock. Photorealistic images. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 6 credits, and the final quote is shown before anything is charged.

Nano Banana 2 is Google’s fast current image-generation and editing option in Gobostock. It accepts long prompts, multiple reference images, a reference video, reproducible seeds, and output presets from 0.5K through 4K, making it useful for iterative commercial concepts as well as high-resolution final images.

Nano Banana Pro is Google’s higher-cost image model for controlled generation and editing. It supports multi-image blending, high-resolution output, and additional reasoning over composition, lighting, and camera direction, so Gobostock positions it for complex production briefs rather than the fastest drafts.

Seedream 5.0 Lite is ByteDance’s efficient image-generation and reference-guided editing model. It is designed for faithful layout, composition, style, and detail control, supports large reference sets and sequential-image workflows, and is offered by Gobostock at 2K and 3K output sizes.

Seedream 5.0 Pro is ByteDance’s flagship generation and editing model for precise production work. It supports up to ten input images, multilingual text rendering, multi-image fusion, and tightly described local changes; Gobostock exposes its premium 1K and 2K generation tiers while keeping claims tied to the controls actually available.

Riverflow 2.5 Standard maps to Sourceful’s Riverflow 2.5 Fast endpoint. It supports text-to-image, image-to-image, editing, custom dimensions, 1K and 2K presets, background controls, prompt enhancement, and selectable thinking levels, giving Gobostock a fast model for design-oriented commercial work with transparent quality controls.

Riverflow 2.5 Pro is Sourceful’s higher-capability image model for commercial production. It supports text-to-image and image-to-image editing with stronger quality and brand control across product, packaging, advertising, and design-heavy workflows; Gobostock exposes 1K, 2K, and 4K output with selectable thinking levels.

GPT Image 2 is OpenAI’s image generation and editing model in the Gobostock catalogue. It supports text-to-image, multi-reference image work, prompt-guided edits, detailed composition, and strong in-image typography, so Gobostock positions it for layouts and scenes where language understanding and readable text are central to the brief.

GPT Image 2.5 Flare is the faster of OpenAI’s 2.5 image models, covering text-to-image, image-to-image, and masked editing. It accepts up to sixteen reference images, renders transparent backgrounds as PNG or WebP, and takes prompts up to 32,000 characters, so it suits briefs that carry a lot of written direction.

GPT Image 2.5 Sunburst is the premium half of OpenAI’s 2.5 pair, tuned for holding a referenced subject steady and keeping visual direction consistent across a set. It shares Flare’s envelope — text-to-image, image-to-image, masked editing, sixteen references, transparency — but is aimed at production work where the same character or product has to survive several images.

Recraft V4.1 is a design-oriented raster image model for illustration, marketing graphics, and brand-consistent artwork. Gobostock exposes it as a prompt-only tool at 1K output: there are no seed, step, guidance, or LoRA controls, so results are driven entirely by the prompt, which can run to about 10,000 characters.

Recraft V4.1 Utility is the fast, lower-cost tier of the Recraft raster family, intended for exploration rather than final art. Gobostock exposes it as a prompt-only tool at 1K output with no seed, step, guidance, or LoRA controls, which makes it the cheapest way to try many directions before committing to a more expensive tier.

Recraft V4.1 Pro is the premium raster tier of the Recraft family, intended for commercial graphics that ship. Gobostock exposes it at 2K output, up to 3072px on the long edge, as a prompt-only tool with no seed, step, guidance, or LoRA controls, so it suits final artwork rather than iterative exploration.

Qwen Image is an image generation model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 6 credits, and the final quote is shown before anything is charged.

Flux Krea is an image generation model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 7 credits, and the final quote is shown before anything is charged.

Hi-Dream Dev is an image generation model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 6 credits, and the final quote is shown before anything is charged.

Hi-Dream Full is an image generation model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 8 credits, and the final quote is shown before anything is charged.

Recraft V4.1 Vector creates scalable SVG artwork through Recraft’s first-party vector-only API. Gobostock expresses palette and background preferences in the prompt, sanitizes the returned markup, and inserts it into Design Studio as a scalable vector group.

Recraft V4.1 Pro Vector is the first-party premium prompt-to-SVG option in Gobostock. It preserves the same sanitized, scalable Design Studio workflow as the Standard tier while targeting more demanding brand and campaign artwork.

Seedream 4.0 Edit is an image editing model from ByteDance, available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 7 credits, and the final quote is shown before anything is charged.

Nano Banana 2 Edit is an image editing model from Google, available in Gobostock. Reference-guided Google editing with up to 14 images. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 63 credits, and the final quote is shown before anything is charged.

Nano Banana Pro Edit is an image editing model from Google, available in Gobostock. Premium Google editing with up to 14 reference images. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 184 credits, and the final quote is shown before anything is charged.

Seedream 5 Lite Edit is an image editing model from ByteDance, available in Gobostock. Reference-guided ByteDance editing at 2K and 3K. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 47 credits, and the final quote is shown before anything is charged.

Seedream 5 Pro Edit is the reference-guided editing configuration for ByteDance Seedream 5 Pro, with premium output tiers and support for multiple references in compatible workflows.

Riverflow 2.5 Standard Edit is an image editing model from Sourceful, available in Gobostock. Riverflow Standard reference editing with up to four images. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 28 credits, and the final quote is shown before anything is charged.

Riverflow 2.5 Pro Edit is an image editing model from Sourceful, available in Gobostock. Premium Riverflow reference editing with up to ten images. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 314 credits, and the final quote is shown before anything is charged.

GPT Image 2 Edit is an image editing model from OpenAI, available in Gobostock. OpenAI editing with up to 16 references and quality tiers. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 8 credits, and the final quote is shown before anything is charged.

GPT Image 2.5 Flare Edit is an image editing model from OpenAI, available in Gobostock. Fast OpenAI editing with masked inpainting and up to 16 references. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 63 credits, and the final quote is shown before anything is charged.

GPT Image 2.5 Sunburst Edit is an image editing model from OpenAI, available in Gobostock. Premium OpenAI editing tuned for reference-subject consistency. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 223 credits, and the final quote is shown before anything is charged.

Flux Kontext Dev is an image editing model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 8 credits, and the final quote is shown before anything is charged.

Flux Kontext Pro is an image editing model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 14 credits, and the final quote is shown before anything is charged.

Flux Kontext Max is an image editing model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 20 credits, and the final quote is shown before anything is charged.

P-Video is Pruna AI's efficient text-to-video and image-to-video model. It supports 720p and 1080p output, 24 or 48 fps, clips from one to ten seconds, optional generated audio, prompt upsampling, and a lower-cost draft mode for iteration.

Seedance 2.0 is ByteDance’s multimodal video model for text, first/last-frame, reference-image, reference-video, and supported audio-guided workflows. It generates synchronized audio, accepts durations from 4 to 15 seconds, and spans 480p through 4K, although Gobostock can generate economically at a lower tier and enhance the approved stock master afterward.

Seedance 2.0 Mini is ByteDance’s faster, lower-cost Seedance variant for text-to-video and image-to-video creation. It supports optional synchronized audio and 4-to-15-second clips at 480p or 720p, making it a practical source model for short stock footage that Gobostock can enhance to a 1080p delivery master.

Seedance 2.5 is ByteDance's production-oriented multimodal video model for longer, reference-heavy commercial work. It supports 4-to-30-second generation at 480p or 720p, image-to-video input, native sound, multilingual prompting, and stronger instruction following.

MiniMax H3 is a multimodal 2K video model that combines text, first/last frames, image, video, and audio references with native synchronized sound. It is designed for consistent subjects, voices, motion, and scene identity across generated or continued shots, and supports 5-to-15-second output in six aspect ratios.

Kling VIDEO 3.0 Standard is Kuaishou’s efficient 720p video model for text-to-video and image-to-video work. It supports 3-to-15-second multi-shot generation, optional native audio, first and last frames, negative prompting, and model-directed scene planning, making it a practical lower-cost source for enhanced 1080p stock clips.

Kling VIDEO 3.0 Pro is Kuaishou’s premium 1080p video model for text-to-video and image-to-video output. It supports 3-to-15-second generation, optional native sound, multi-shot scene planning, first and last frames, and negative prompts, so Gobostock can use the native file directly as a full-HD stock master without a mandatory upscale pass.

Kling VIDEO 3.0 4K is Kuaishou’s native 4K generation tier for text-to-video and image-to-video clips. It supports multi-shot output, optional native audio, and 3-to-15-second durations without an external upscaling workaround, so it is best reserved for stock footage where true high-resolution generation justifies the higher cost.

Gemini Omni Flash 1.1 is Google’s multimodal video model, generating 3-to-10-second clips with native synchronized audio from text or a still image. Over the previous release it adds 1080p and 4K output, a cheap 360p draft mode for blocking a shot out, scene extension, and frame interpolation for stretching a clip toward 30 seconds.

FLUX 3 Video is Black Forest Labs’ video model, extending the FLUX family from stills into motion. It generates from text or a still image, carries synchronized audio by default, and accepts keyframes to pin how a shot begins and ends. Clips run from 5 to 20 seconds at 720p or 1080p across six aspect ratios, including a 21:9 cinematic crop.

Wan 3.0 is Alibaba’s all-in-one multimodal video model, built for longer and more reference-heavy work than its predecessors. It handles text-to-video, first- and last-frame guidance, and reference-driven generation in one model, produces native audio, and supports clips up to 30 seconds — the longest single generation in the Gobostock catalogue — at 480p, 720p, or 1080p.

P-Video 2 is Pruna AI’s second-generation video model, tuned for speed and cost rather than maximum fidelity. It generates from text or a still image with optional audio, supports first- and last-frame guidance, and runs at 24 or 48 frames per second. It is the least expensive per second of any video model in the Gobostock catalogue, which makes it the natural choice for iteration.

MiniMax H3 Max is the top tier of the H3 family, generating from text or a still image at 480p or 768p with reference-audio support. Clips run from 5 to 15 seconds, and prompt expansion can be left off, balanced, or pushed toward quality. It is materially cheaper per second than the standard H3 while keeping the same reference-driven approach.

Kling Omni 3.0 Standard is a video generation model from Kling AI, available in Gobostock. Per-second base price at 720p for Omni edits / replace fallback. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 134 credits, and the final quote is shown before anything is charged.

Kling Omni 3.0 Pro is a video generation model from Kling AI, available in Gobostock. Per-second base price at 1080p for Omni edits / replace fallback. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 187 credits, and the final quote is shown before anything is charged.

HappyHorse 1.1 is Alibaba’s expressive audio-visual generation model for text and image-driven clips. Gobostock exposes always-on native sound, reproducible seeds, 720p or 1080p output, first-frame guidance, multiple reference images, and 3-to-15-second durations for short cinematic and character-focused footage.

Fish Audio S2.1 Pro is a multilingual text-to-speech model for expressive, low-latency narration and dialogue. It supports more than 80 languages, automatic language detection, inline emotion and delivery cues, multi-speaker dialogue, voice-library choices, and supported reference-voice cloning.

Gemini 3.1 Flash TTS is Google’s preview text-to-speech model for controllable audio from exact text. It detects the spoken language automatically across more than 70 languages, and Google documents single-speaker and two-speaker output, natural-language direction for style, accent, pace, and tone, and 30 prebuilt voices; Gobostock keeps each voice tied to the engine it belongs to.

Clone Voice uses Alibaba’s Qwen3-TTS 1.7B Base model to synthesize multilingual speech from a consented reference recording. It supports a short reference, optional matching transcript, language selection, playback speed, and higher-similarity in-context mode, while Gobostock keeps every saved clone private and tied to the engine that created it.

Describe Voice uses Alibaba’s Qwen3-TTS 1.7B VoiceDesign model to create speech from a written voice description and sample script. Gobostock keeps the resulting voice private and tied to the engine that created it, and uses the model for designed-voice workflows where creators need a new vocal character instead of imitating a real person.

Mirelo SFX is Gobostock’s live prompt-driven sound-effects engine, generating short layered sound design from a text description. It suits UI clicks, transitions, ambiences, and impact effects where a bespoke cue is faster to describe than to find in a library.

Sonilo SFX 1.1 is a text-to-sound-effects engine available in Gobostock. It generates commercial-use-safe effects from a prompt, with control over duration and audio format up to 30 seconds.

Sonilo Video SFX 1.1 is a video-to-sound-effects engine available in Gobostock. Instead of taking a written description, it watches the clip you give it and generates foley and ambience timed to what happens on screen — footsteps landing where feet land, a door closing when the door closes. A prompt is optional and steers the result rather than defining it. Gobostock can return the generated track on its own or the original clip with the sound already mixed in, at the same price either way.

Sonilo Video Music 1.1 is a video-to-music engine available in Gobostock. It watches a clip and composes a score that follows it, rather than generating music from a text description alone. An optional prompt sets the mood and a prompt-influence control decides how strongly that prompt competes with what the model sees. Unlike the sound-effects model, the length of the score is chosen rather than inherited from the clip, so a cue can be shorter or longer than the footage it was written for.

Background Removal is an utility model available in Gobostock. Remove image backgrounds. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 1 credits, and the final quote is shown before anything is charged.

Qwen Image Layered decomposes one raster source into a layered TIFF response. Gobostock extracts the pages as aligned transparent PNG objects and verifies their recomposition before importing them into Design Studio.

Seedream 5 Pro Layerize splits one raster image into a rebuilt background plus independently editable transparent PNG layers. Design Studio presents this as Layerize Pro.

Recraft Vectorize converts one eligible raster source into SVG. Gobostock uses the original entitled source, sanitizes the returned markup, and imports the result as a scalable vector group.

Upscale (Clarity) is an utility model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 11 credits, and the final quote is shown before anything is charged.

Upscale (CCSR) is an utility model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 14 credits, and the final quote is shown before anything is charged.

Upscale (SwinIR) is an utility model available in Gobostock. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 8 credits, and the final quote is shown before anything is charged.

Upscale (Default) is an utility model available in Gobostock. Default upscale model (2x). It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 10 credits, and the final quote is shown before anything is charged.

Upscale (Recraft Crisp) is an utility model available in Gobostock. Recraft Crisp upscale up to 4K — flat per image. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 6 credits, and the final quote is shown before anything is charged.

Upscale (SeedVR2) is an utility model available in Gobostock. SeedVR2 upscale — priced per output megapixel. It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 2 credits, and the final quote is shown before anything is charged.

Video Enhance 1080p is an utility model from ByteDance, available in Gobostock. ByteDance video upscaler to 1080p (per second). It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 12 credits, and the final quote is shown before anything is charged.

Video Enhance 2K is an utility model from ByteDance, available in Gobostock. ByteDance video upscaler to 2K (per second). It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 24 credits, and the final quote is shown before anything is charged.

Video Enhance 4K is an utility model from ByteDance, available in Gobostock. ByteDance video upscaler to 4K (per second). It runs inside the Gobostock creation workflow, where the controls on offer, the accepted inputs, and the available output sizes are set by the tool rather than configured by hand. Generations start from 48 credits, and the final quote is shown before anything is charged.