Choose the model. Keep the same API.

Strong multilingual text, structured visual layouts, reference fusion, and precise natural-language edits.

High-control image generation and editing for posters, products, infographics, and iterative design work.

Commercial-grade visuals with strong composition, detail, reference consistency, and high-resolution output.

Context-aware image editing that preserves subjects and style across targeted changes.

Fast text-to-image generation with readable typography, flexible ratios, and detailed output.

A balanced route for generation, local editing, multiple references, and production image workflows.

Distinctive art direction, polished aesthetics, and reference-driven visual exploration.

PixMind's free image generation route for quick everyday text-to-image creation.

A cost-efficient Nano Banana Pro route covering text-to-image and image editing with reference support.

Google's Nano Banana 2 high-speed image generation across 1K, 2K, and 4K resolutions.

The original Nano Banana image generation route for versatile text-to-image and editing work.

An economical Nano Banana 2 variant balancing speed, quality, and cost for everyday generation.

Midjourney's anime and illustration-tuned model for stylized characters and vibrant artwork.

Midjourney V8.2 with refined aesthetics, stronger prompt adherence, and richer art direction.

Midjourney V8.1 delivering elevated visual quality and consistency for polished creative output.

ByteDance Seedream 5.0 commercial-grade image generation with strong composition and detail.

Seedream 4.5 balanced image generation with reliable quality for everyday creative work.

Seedream 4.0 image generation covering text-to-image and editing for production visuals.

GPT Image 1.5 high-control generation and editing for design-heavy visuals.

GPT Image 4o versatile image generation with precise prompt following and editing.

An economical GPT Image 2 variant for cost-aware generation and editing at scale.

Wan 2.6 image generation and editing from Alibaba for multilingual creative workflows.

Wan 2.7 Image Pro, Alibaba's higher-tier route for sharper, production-ready visuals.

Qwen Image 3.0 Pro with improved text rendering and high-resolution output.

Qwen Image 2.0 Pro text-to-image generation with readable typography and detail.

xAI image generation for expressive, high-impact visuals across seven common aspect ratios and larger batches.

Fast Krea text-to-image generation with flexible aspect ratios, deterministic seeds, and 1K or 2K output.

Grok Imagine 1.5 image generation with bold, expressive visual styling.

Recraft V4 vector generation for clean, scalable, print-ready artwork.

Recraft V4 Pro vector generation with premium quality for professional design.

Recraft image-to-vector conversion that turns raster images into editable vectors.

Cinematic video generation with image guidance, shot control, first/last frames, and native audio.

Native 2K video with synchronized audio from text, images, first/last frames, and multimodal references.

Multi-reference video creation with coordinated characters, scenes, motion, and sound.

Native 30-second single-shot video with up to 50 multimodal references, 4K output, and region-level editing.

High-fidelity image animation and camera or subject motion control for production shots.

Versatile text, image, reference, and first/last-frame workflows through one video route.

Wan 3.0 video generation with multimodal references, synchronized audio, 2-30 second clips, and up to 1080p output.

Detailed video generation for longer creative shots with image guidance and audio-capable output.

Fast video ideation across text-to-video, image animation, transitions, effects, and sound.

Reference-led character performance and expressive motion for short-form creative video.

Kuaishou Kling 3.0 Turbo, a fast text-to-video and image-to-video route for quick drafts.

ByteDance Seedance 1.5 Pro video generation with audio-synced output.

Seedance 2.0 Fast, a quicker Seedance route for faster video turnaround.

Seedance 2.0 Mini, a lightweight variant for efficient short video generation.

OpenAI Sora 2 video generation for text-to-video and image-to-video creative shots.

An economical Sora 2 variant for cost-efficient video generation.

Google Veo 3 cinematic video generation with image guidance and audio.

Veo 3.0 Fast, a quicker Veo route for rapid video iteration.

An economical Veo 3.1 variant balancing quality and cost.

Veo 3.1 Fast Eco, the most cost-optimized Veo route for fast, budget-friendly video.

PixVerse V6 video generation with text-to-video, image animation, and effects.

Wan 2.6 image-to-video from Alibaba for animating still images.

Wan 2.6 text-to-video from Alibaba for turning prompts into motion.

Wan 2.6 I2V Flash, a rapid image-to-video variant for quick animation.

Happy Horse 1.1, an upgraded route for reference-led character performance video.

Anthropic's Mythos-class route for autonomous knowledge work, long-form reasoning, and coding. Choose Fable 5 when an agent must retain context across extended tasks, inspect visual inputs, and coordinate tools before producing a concise result.

Anthropic's flagship route for demanding reasoning, end-to-end software work, visual analysis, and autonomous tools. It fits high-stakes workflows where solution quality and persistence matter more than choosing the lowest-cost model.

A balanced Claude route for production coding, agents, professional writing, and visual reasoning. Sonnet 5 is the practical default when you need strong multi-step work with lower cost than the Opus tier.

A high-capability Opus route for complex reasoning, coding, vision, and tool-driven execution. Use it for long tasks that benefit from careful planning, repository context, and consistent decisions across many steps.

A cost-optimized Claude Opus 4.8 variant focused on coding and tool-driven workloads. It is useful when you want Opus-class task handling while reducing the entry token cost for repeated engineering runs.

Planned multimodal route for text, vision, reasoning, streaming, and tool use.

Planned reasoning-focused route for technical analysis, coding, and streamed responses.

Higher-capability DeepSeek V4 text route for complex reasoning, software architecture, careful code review, and long-document analysis with a one-million-token context.

Cost-efficient DeepSeek V4 text route for high-volume extraction, classification, rewriting, summaries, and lightweight coding with a one-million-token context.

Zhipu AI's large reasoning route for complex software engineering and long-horizon agents. Its long-context design suits repository-scale analysis, multi-step tool use, and streamed technical work where earlier decisions must remain consistent.

A cost-efficient multimodal GLM route for focused coding, visual reasoning, and agent subtasks. Choose the Flash tier for high-volume tool workflows that need long-context behavior without the full GLM-5.3 entry cost.

The flagship GPT-5.6 route for complex reasoning, coding, vision, and agentic execution. Sol is the quality-first choice for command-line work, multi-step engineering, and tool workflows that can justify its higher token price.

The balanced GPT-5.6 tier for everyday coding, reasoning, visual input, and tools. Terra sits between Sol and Luna when a production workflow needs capable agent behavior without paying for the flagship route on every request.

The lightweight GPT-5.6 tier for high-volume, latency-sensitive text work. Use Luna for chat, extraction, classification, and simple streamed workflows where cost efficiency matters more than deep multimodal reasoning.

Google's frontier preview route for complex reasoning, software engineering, and multimodal analysis. It accepts text, images, and video through PixMind and is best reserved for difficult workflows that need Pro-tier depth and tool use.

A high-efficiency Gemini model that brings strong coding, reasoning, and multimodal understanding to Flash-tier workloads. It fits responsive agents and parallel processing where throughput and capable visual analysis must stay balanced.

A responsive multimodal route for coding, agent workflows, and polished web or application output. Choose it for production tasks that need stronger reasoning and visual context while retaining Flash-tier efficiency.

Google's most capable Flash-tier route for long-horizon software engineering, autonomous agents, and complex multi-step workflows. Through PixMind it combines text, image, audio, video, and PDF input with reasoning, tools, coding, and streamed text output.

Google's multimodal Flash route for fast agentic workflows, coding, and reliable multi-step reasoning. It is a practical option for responsive products that combine text, image, or video context with streamed output.

Google's most cost-efficient Flash tier for focused, repeatable subtasks. Use it for high-volume chat, extraction, classification, and lightweight agents where a smaller price footprint matters more than maximum reasoning depth.