Modele AI

Most capable Opus — enhanced coding, agentic workflows, and long-horizon reasoning with 1M context

Previous Opus — enhanced SWE, vision, and long-horizon agentic reasoning with 1M context

Hybrid reasoning model with 1M context, top-tier coding and agentic performance

Previous premium model with maximum intelligence

Best combination of speed and intelligence, near-flagship performance

Previous smart model for complex agents and coding

Fastest model with near-frontier intelligence

Newest frontier model — highest reasoning for coding and professional work

Highly capable GPT model for coding and agentic tasks

Most powerful GPT model with maximum compute for complex reasoning

The best model for coding and agentic tasks across industries

Intelligent reasoning model with configurable reasoning effort

Previous intelligent reasoning model for coding and agentic tasks

A faster, cost-efficient version of GPT-5

Fastest, most cost-efficient version of GPT-5

Smarter and more precise responses

Smartest non-reasoning model

Smaller, faster version of GPT-4.1

Reasoning model for complex tasks

Version of o3 with more compute for better responses

Fast, cost-efficient reasoning model

Most advanced reasoning model with complex problem-solving

Cheapest frontier-class model — half the cost of Gemini 3 Flash with strong tool calling

Low-latency Live API model for real-time dialogue and voice-first AI applications

Frontier intelligence optimized for agentic workflows, coding, and video at higher speed

Frontier intelligence with superior search and grounding

State-of-the-art thinking model for complex problems

Best price-performance for large scale processing

Fastest flash model for cost-efficiency

Fastest, most intelligent Grok — 1M context, 3 reasoning levels, top agentic tool calling

Latest Grok beta optimized for multi-agent orchestration

Latest Grok beta with extended reasoning

Fast Grok without reasoning overhead

Smaller, faster Grok with reasoning

1M context, thinking + non-thinking modes, tool calls

Flagship model — 1M context, thinking + non-thinking modes

Alibaba's flagship general-purpose LLM via DashScope - top-tier reasoning and coding

Automated deep research - plans research steps, performs web searches, generates structured reports

Recursive self-improvement — SOTA in software engineering, tool calling, and office productivity

M2.7 at ~100 tps — same performance, faster and more agile

Peak performance and ultimate value — master the complex

M2.5 at ~100 tps — same performance, faster and more agile

Polyglot programming mastery with precision code refactoring

Agentic capabilities with function calling and advanced reasoning

ByteDance flagship — 76.5% SWE-Bench, 98.3% AIME 2025, hour-long video understanding

Meta's flagship Llama 4 model

Efficient Llama 4 model

Meta's powerful open source model

Meta's powerful open source model

Balanced Mistral model

Latest large Mistral model

Creative writing focused model

Large Qwen model with 235B parameters

Vision-language Qwen model

Most powerful Qwen model

Latest Z-AI flagship — enhanced long-horizon coding and autonomous agent tasks

Z-AI flagship model with strong reasoning and tool use

Fast inference model optimized for agentic workflows and tool use

Latest GLM model

Powerful GLM model

Lightweight GLM model

Xiaomi's 1T-parameter flagship — agentic workflows, tool calling, and advanced reasoning with 1M context

Xiaomi's fast AI model

Nvidia's compact model

Long-horizon coding, UI/UX generation, and multi-agent orchestration with parallel sub-agents

State-of-the-art visual coding and agentic tool-calling with multimodal reasoning

196B MoE reasoning model — activates 11B per token, extremely fast

Hybrid attention + MoE vision-language model with 1M context

1T parameter frontier model built for agentic multi-step reasoning

Omni-modal frontier model with vision, hearing, reasoning, and action

Flagship uncensored reasoning model from Nous Research — hybrid think/respond mode, low refusal rates, strong at math, code, and structured output

Efficient uncensored reasoning model from Nous Research — hybrid think/respond mode, low refusal rates, strong at math, code, and structured output

Versatile multimodal model with low latency for agent and vision tasks

Kunya V1

Kunya

Premium

Intelligently routed model — Opus-level quality at budget cost. Routes to the best model for each request.

High-efficiency image generation optimized for speed and volume, up to 4K with thinking

Professional asset production with advanced reasoning and 4K output

Fast native image generation with editing — the original Gemini image model

Latest state-of-the-art image generation with fast, high-quality output and flexible sizes

Image generation with native editing

Image generation with native editing support

High quality image generation with text rendering

Fast and affordable image generation

Top-tier image quality with editing and multi-reference support

High-end image generation with strong prompt adherence and editing

Complex text, typography, and multi-reference editing

Fastest and most cost-effective FLUX model

ByteDance model with strong editing consistency and text rendering

Most powerful Riverflow with perfect text rendering and 4K support

Fastest Riverflow for production and latency-critical workflows

Most powerful Riverflow V2 preview - unified text-to-image and image-to-image

Standard Riverflow V2 preview with great quality

Fastest Riverflow V2 preview model

Ultra-fast image generation in ~1 second

Latest SD with improved quality, typography, and prompt understanding

Fast SD 3.5 Large with 4-step generation

Balanced SD 3.5 with great quality/speed ratio

Stable Diffusion XL - high quality 1024x1024 images

Ultra-fast SDXL with 4-step generation

SDXL with customizable LoRA weights for fine-tuned styles

High-quality image generation with LoRA fine-tuning support

Qwen's native image generation model

Professional-grade image generation with clean licensing

Advanced image editing with precise control

Open-source flow-based image generation

High-quality bilingual image generation (English/Chinese)

ByteDance Seedream 5.0 Lite image editing — intelligent multi-image editing with reasoning, style transfer, and beautification (2K-3K)

Alibaba's flagship image generation - high realism, fine detail, excellent text rendering

Alibaba's image editing model - modify text, add/remove objects, style transfer, detail enhancement

Lightweight fast image generation with Chinese & English text rendering

Alibaba Wan 2.6 text-to-image generation - photorealistic to illustrative styles

ByteDance Seedream 5.0 Lite — high-quality 2K/3K image generation with text-to-image and image editing

Midjourney V7 — industry-leading image generation with stunning aesthetics. 4 images per generation. Supports --ar, --s, --c and all V7 parameters.

ByteDance Seedream 5.0 via Dreamina/ModelArk — high-quality 2K image generation. Admin-only for comparison with Evolink provider.

ByteDance Seedream 5.0 Lite via Dreamina/ModelArk — fast 2K image generation. Admin-only for comparison with Evolink provider.

Kunya V1 Image

Kunya

Intelligently routed image generation — Z-Image Turbo for fast/cheap, Seedream for quality, GPT Image for editing.

OpenAI Sora 2 — physics-aware world simulation with audio (up to 12s, 720p)

OpenAI Sora 2 Pro — highest quality with audio (up to 12s, 1080p)

OpenAI Sora 2 — animate images with physics simulation (up to 12s, 720p)

OpenAI Sora 2 Pro — highest quality image animation (up to 12s, 1080p)

OpenAI Sora 2 — transform existing videos with style changes

Google Veo 3.1 — fast cinematic generation (up to 8s, 720p)

Google Veo 3.1 — cinematic video (up to 8s, 1080p)

Google Veo 3.1 — image-to-cinema (up to 8s, 1080p)

Google Veo 3.1 Extend — continue an existing video up to ~30s total (720p/1080p)

Google Veo 3.1 — generate video from up to 3 reference images (up to 8s, 1080p)

Google Veo 3.1 — animate between a first and last keyframe (up to 8s, 1080p)

AI video generation from text, images, and video with native audio

High-quality video with excellent character consistency

Transform images into videos with motion

Professional video generation

Professional image-to-video generation

Photorealistic video with incredible motion (5s or 9s)

Fast version of Ray 2 for quicker generation (5s or 9s)

Realistic motion and physics-aware generation (5s)

Fast cinematic video from images (5s or 10s, 768p)

Narrative-coherent video (fixed 6s clips; use scene chaining for longer)

Real-time video generation (fixed 6s clips)

Latest MiniMax model — cinematic motion, expressive faces, anime & illustration styles, 15 camera commands

Fast & cost-effective image-to-video — same quality, optimized for speed

Tencent open-source video model

High-quality text-to-video generation

Transform images into dynamic videos

Open-source video generation model

Affordable high-quality video generation

Enhance video resolution and quality

Increase video frame rate smoothly

Open-source model with 20s 4K support and improved quality

Animate images with LTX v2 - up to 20 seconds

Basic face swap in images and videos

Premium face swap with hair preservation, 2x upscale, and detail enhancement

Swap faces on GIFs — fun for social sharing

Make any portrait mimic your expressions - face puppeteering

Fast face puppeteering - your expressions control any face

ByteDance motion transfer — full body, expressions, lip movement from driving video to any character (humans, animals, cartoons)

ByteDance OmniHuman — audio-driven avatar animation with emotion and cognitive simulation

ByteDance OmniHuman 1.5 — film-grade talking avatar from photo + audio with micro-expressions and cognitive simulation

Most powerful lipsync — native visual intelligence for professional-quality video-to-video

High-quality realistic lipsync preserving natural teeth and unique facial features

Realistic lipsync animations from audio with advanced synchronization

Budget-friendly video-to-video lip sync — $0.20 flat for up to 40s, then $0.005/s

Kling audio-to-video lip sync — realistic lip movements from audio (2-60s audio, 720p/1080p)

Portrait animation with audio-driven lip sync

Lip sync video generation from audio input — up to 60s

Real-time lip sync for virtual presenters — up to 120s

Wan 2.2 A14B — high-quality anime/artistic video with improved motion and expressions (480p-720p)

Wan 2.2 motion transfer — replicate expressions and movements from a reference video onto a character image

Wan 2.2 character replacement — replace the character in a video while preserving scene lighting and motion

Transform videos with anime and artistic styles

Anime-style video with motion control from sparse frames

Anime and artistic video generation (superseded by Wan 2.2)

Image-to-anime animations (superseded by Wan 2.2)

Generate cartoon/anime interpolation between keyframes

Kling face puppeteering - drive faces with your video

Kling v2.5 lip sync — superseded by Kling LipSync audio-to-video endpoint

Kling O3 Pro — generate the next shot from a reference video, preserving motion & camera style (3-15s, 1080p)

Kling O3 Standard — generate the next shot from a reference video (3-15s, 720p)

Kling O3 Pro — edit existing videos with element injection and style transfer (3-15s, 1080p)

Kling V3 Pro — cinematic text-to-video with multi-shot and native audio (3-15s, 1080p)

Kling V3 Pro — animate images with multi-shot storyboarding (3-15s, 1080p)

Kling O3 Standard — text-to-video with multi-shot and audio (3-15s, 720p)

Kling O3 Standard — animate images with start/end frame control (3-15s, 720p)

Kling O3 Standard — reference-to-video with @Element character locking + @Image style refs (3-15s, 720p)

Kling O3 Pro — reference-driven text-to-video with character consistency (3-15s, 1080p)

Kling O3 Pro — best-in-class image-to-video with element referencing (3-15s, 1080p)

Kling O3 Pro — reference-to-video with @Element character locking (frontal+multi-angle refs) + @Image style refs (3-15s, 1080p)

Kling O3 4K — reference-to-video with @Element character locking at native 4K. Up to 7 refs (3-15s)

Kling V3 Native 4K — professional-grade 4K video from text (3-15s)

Kling V3 Native 4K — professional-grade 4K video from images (3-15s)

Kling O3 Native 4K — professional-grade 4K video with reference support (3-15s)

Kling O3 Native 4K — professional-grade 4K video from images with references (3-15s)

Kling V3 Standard via direct API — 720p text-to-video (5/10/15s)

Kling V3 Pro via direct API — 1080p text-to-video (5/10/15s)

Kling V3 Standard via direct API — 720p image-to-video (5/10s)

Kling V3 Pro via direct API — 1080p image-to-video (5/10s)

Kling O3 Standard via direct API — 720p text-to-video (3-15s)

Kling O3 Pro via direct API — 1080p text-to-video (3-15s)

Kling O3 Standard via direct API — 720p image-to-video (3-15s)

Kling O3 Pro via direct API — 1080p image-to-video (3-15s)

Kling V3 native 4K text-to-video via direct API (3-15s)

Kling V3 native 4K image-to-video via direct API (3-10s)

Kling O3 native 4K text-to-video via direct API (3-15s)

Kling O3 native 4K image-to-video via direct API (3-15s)

ByteDance Seedance 2.0 via FAL — cinematic T2V with native audio, up to 15s at 1080p

ByteDance Seedance 2.0 via FAL — animate images with native audio, start/end frame control, up to 15s

ByteDance Seedance 2.0 via FAL — multimodal ref system: up to 9 images + 3 videos + 3 audio, native audio

ByteDance Seedance 2.0 Fast via FAL — lower latency and cost, up to 15s

ByteDance Seedance 2.0 Fast via FAL — fast image-to-video with native audio

ByteDance Seedance 2.0 Fast via FAL — fast multimodal reference, up to 9 images + 3 videos + 3 audio

Alibaba Happy Horse 1.0 — #1 ranked AI video model, native audio + lip-sync, up to 15s at 1080p

Alibaba Happy Horse 1.0 — #1 ranked I2V with native audio, multilingual lip-sync, up to 15s at 1080p

Alibaba Happy Horse 1.0 — reference-driven video with character consistency (1-9 images), native audio, 1080p

Alibaba Happy Horse 1.0 — natural language video editing with up to 5 reference images, 1080p

Alibaba Wan 2.6 - image-to-video with audio, up to 15s at 1080p

Alibaba Wan 2.6 - higher quality image-to-video, up to 15s at 1080p

Alibaba Wan 2.6 - cinematic multi-shot text-to-video with audio, up to 15s at 1080p

Alibaba Wan 2.2 - generate video from first and last frame images, 5s at 1080p

Alibaba Wan 2.6 - replicate character appearance from reference videos, multi-character support, up to 10s

Alibaba Wan 2.1 - multi-image reference, video redraw, local editing, extension, frame expansion

Alibaba Wan 2.2 - animate a person image using motion from a reference video, up to 30s

Alibaba Wan 2.2 - replace people in videos with people from images, keeping original background, up to 30s

ByteDance Seedance 2.0 — text-driven video with synchronized audio, lip-sync, web search, up to 15s

ByteDance Seedance 2.0 — first/last frame image-driven video with synchronized audio, up to 15s

ByteDance Seedance 2.0 — multimodal @-reference system: up to 9 images + 3 videos + 3 audio tracks

ByteDance Seedance 2.0 Fast — faster text-driven video at lower cost, synchronized audio, up to 15s

ByteDance Seedance 2.0 Fast — faster image-driven video at lower cost, synchronized audio, up to 15s

ByteDance Seedance 2.0 Fast — faster multimodal @-reference at lower cost, up to 9 images + 3 videos + 3 audio

ByteDance Seedance 1.5 — synchronized audio+video generation with lip-sync and foley (up to 12s)

Kling V3 — standard text-to-video with multi-shot and sound effects (5s or 10s)

Kling V3 — image-to-video with first/last frame, multi-shot, and sound effects (5s or 10s)

Kling V3 — motion transfer from reference video to character in reference image (up to 10s per render)

Kling O3 (V3 Omni) — highest quality text-to-video with multi-shot and sound (3-15s)

Kling O3 (V3 Omni) — best-in-class image-to-video with reference images, elements, and multi-shot (3-15s)

Alibaba Wan 2.7 — multi-shot narrative, auto BGM/SFX or driving-audio lip-sync, 2-15s

Alibaba Happy Horse 1.0 — #1 ranked text-to-video, native audio + lip-sync, 3-15s

Alibaba Happy Horse 1.0 — image-to-video with native audio, 3-15s

Alibaba Happy Horse 1.0 — reference-driven video with 1-9 images, native audio, 3-15s

Alibaba Happy Horse 1.0 — natural language video editing with up to 5 reference images

Kling O1 — style-focused image-to-video with first/last frame support (5s or 10s)

Kunya V1 Video

Kunya

Intelligently routed video generation — Kling for quality, Seedance for speed, resolution-aware selection.

Speech-to-text transcription

Text-to-speech optimized for speed

Text-to-speech optimized for quality

Powerful, low-latency speech generation with expressive audio tags for precise narration control — 70+ languages

Google Cloud Text-to-Speech — standard voices, 40+ languages

Google Neural2 voices — highly natural-sounding TTS using novel synthesis methods

Google's most expressive TTS — Chirp3 HD voices with studio-quality audio

Google Studio voices — highest quality, human-like expressiveness

ElevenLabs Eleven v3 — ultra-realistic voice synthesis with 30+ languages and voice cloning

ElevenLabs Flash v2.5 — lowest latency TTS for real-time applications, 32 languages

Alibaba's multilingual TTS with 49 voices, 10+ languages - ElevenLabs alternative

Snapshot version of Qwen3 TTS Flash with 49 voices

Instruction-controllable TTS - control speech style via text instructions, 10+ languages

Generate custom voices from text descriptions - design unique voices without audio samples

Clone voices from 10-20 second audio samples - highly natural voice replication

Next-gen generative TTS model - high-quality real-time streaming synthesis

Fast CosyVoice TTS - cost-effective streaming synthesis

Google DeepMind real-time streaming music generation with interactive steering

Generate music from text prompts with optional reference audio

Lyric-driven composition with synchronized vocals and structure tags

Ultra-fast professional music generation - 3 min track in under 10s

Full songs in any style with lyrics, tags, and BPM control

Royalty-free instrumental music with stem generation for remixing

High-quality music and sound design generation

Meta's large music generation model

Studio-grade music with vocals or instrumentals, up to 10 min, multilingual lyrics

Latest Suno model — superior musical expression, fast generation, vocals + instrumentals

Recommended Suno model — smarter prompts, up to 8 min, great vocal quality

Enhanced V4.5 with richer tones and new creative methods, up to 8 min

V4.5 full-featured — all capabilities unlocked, up to 8 min

Improved vocal quality, up to 4 min, lighter and faster generation

Most capable agentic coding model with frontier reasoning

Optimized for agentic coding

Most intelligent coding model for long-horizon tasks

Optimized for agentic coding in Codex

Alibaba's flagship code model via DashScope - code generation, completion, and debugging

Fast, cost-effective code model via DashScope for rapid code tasks

123B agentic coding model with 256K context

Fast coding model for completion, correction, and test generation

123B agentic coding model

Enhanced coding model

Fast coding model

Coding assistant from Kwaipilot