Magai includes chat, image, and video models in one list. You can switch models at any time in a chat. Every Magai plan includes all of these models. What changes by plan is how much monthly usage you get.
How to choose a model
In a chat, open the model picker. Search by name, or sort by Model, Context, or Usage. You can filter the list by All, Moderated, or Unmoderated, and by capability (for example Vision, Reasoning, Text to Image, or Text to Video).
New chats start on Claude Sonnet 5. Choose Auto if you want Magai to pick a model for you.
The picker always shows the current models, context limits, and usage. If a model is added or updated, it will appear there first.
How usage works
Each model has a usage multiplier. A lower number uses less of your monthly usage. A higher number uses more. For example, Claude Sonnet 5 is 0.8x, and GPT-5.5 Pro is 14x.
Auto does not show a usage multiplier in the picker.
Standard includes 1x usage, Pro includes 3x, and Ultra includes 15x. You can also top up without changing plans. See How Much Usage Does Each Plan Include?
All models
Model | Context | Usage |
GLM 5.3 | 1M | 0.39x |
Gemini 3.7 Flash | 1M | 0.3x |
Grok 4.6 | 500K | 0.53x |
Grok Imagine Image | N/A | 0.35x |
Hailuo | N/A | 4.5x |
Seedance | N/A | 0.9x |
Auto | 128K | — |
Claude Haiku 4.5 | 200K | 0.4x |
Claude Opus 4.8 | 1M | 2x |
Claude Opus 5 | 1M | 2x |
Claude Sonnet 5 | 1M | 0.8x |
DeepSeek v4 Flash | 1M | 0.018x |
DeepSeek v4 Pro | 1M | 0.087x |
Flux | N/A | 1x |
Gemini 2.5 Flash | 1M | 0.2x |
Gemini 2.5 Pro | 1M | 0.8x |
Gemini 3.1 Flash Lite | 1M | 0.3x |
Gemini 3.1 Pro | 1M | 2x |
Gemini 3.5 Flash | 1M | 0.7x |
Gemini 3.5 Flash-Lite | 1M | 0.19x |
Gemini 3.6 Flash | 1M | 0.6x |
Gemini Omni Flash | N/A | 6.2x |
GLM 5 | 203K | 0.22x |
GLM 5 Turbo | 203K | 0.28x |
GLM 5.2 | 1M | 0.27x |
GPT Image | N/A | 5x |
GPT OSS 120B | 131K | 0.02x |
GPT-5.4 Mini | 400K | 0.35x |
GPT-5.4 Nano | 400K | 0.1x |
GPT-5.5 | 1M | 2.4x |
GPT-5.5 Pro | 1M | 14x |
GPT-5.6 Luna | 1M | 0.09x |
GPT-5.6 Luna Pro | 1M | 0.09x |
GPT-5.6 Sol | 1M | 2.33x |
GPT-5.6 Sol Pro | 1M | 2.33x |
GPT-5.6 Terra | 1M | 0.93x |
GPT-5.6 Terra Pro | 1M | 0.93x |
Grok 4.5 | 500K | 0.53x |
Grok Imagine Video | N/A | 3.3x |
Ideogram | N/A | 2x |
Kimi K2.5 | 262K | 0.23x |
Kimi K3 | 1M | 1.2x |
Kling Image | N/A | 1x |
Kling Video | N/A | 7x |
Leonardo Motion | N/A | 1.1x |
Leonardo.ai | N/A | 2x |
Llama 4 Maverick | 1M | 0.05x |
Llama 4 Scout | 328K | 0.03x |
Luma | N/A | 8.3x |
Luma UNI-1 | N/A | 12x |
MiMo v2 Omni | 262K | 0.16x |
MiMo v2 Pro | 1M | 0.27x |
MiniMax | N/A | 8.3x |
MiniMax M2.7 | 205K | 0.1x |
MiniMax M3 | 1M | 0.2x |
Mistral Large 3 | 262K | 0.13x |
Mistral Pixtral | 128K | 0.6x |
Mistral Small 4 | 262K | 0.05x |
Nano Banana | N/A | 1x |
Nemotron 3 Nano | 262K | 0.02x |
Nova 2 Lite | 1M | 0.19x |
Nova Pro | 300K | 0.3x |
o4 Mini | 200K | 0.4x |
Perplexity Deep Research | 200K | 0.7x |
Perplexity Sonar | 127K | 0.2x |
Perplexity Sonar Pro | 200K | 1.2x |
Perplexity Sonar Pro Search | 200K | 1.2x |
Recraft | N/A | 1x |
Reve | N/A | 1x |
Runway Image | N/A | 1x |
Runway Video | N/A | 4.2x |
Seedream | N/A | 1x |
Veo | N/A | 13.3x |
Voice GPT | 32K | 1x |
Image models
Use the capability filters in the picker to find image models.
Text to Image: Grok Imagine Image (0.35x), Flux (1x), GPT Image (5x), Ideogram (2x), Leonardo.ai (2x), Luma UNI-1 (12x), Nano Banana (1x), Recraft (1x), Reve (1x), Seedream (1x)
Image to Image: Grok Imagine Image (0.35x), Flux (1x), GPT Image (5x), Leonardo.ai (2x), Luma UNI-1 (12x), Nano Banana (1x), Recraft (1x), Reve (1x), Runway Image (1x), Seedream (1x)
Multi Images to Image: Flux (1x), GPT Image (5x), Nano Banana (1x), Reve (1x)
Edit Mode: Grok Imagine Image (0.35x), Flux (1x), GPT Image (5x), Grok Imagine Video (3.3x), Nano Banana (1x), Seedream (1x)
Kling Image is also in the model list (1x).
Video models
Text to Video: Hailuo (4.5x), Seedance (0.9x), Gemini Omni Flash (6.2x), Grok Imagine Video (3.3x), Kling Video (7x), Leonardo Motion (1.1x), Luma (8.3x), MiniMax (8.3x), Veo (13.3x)
Image to Video: Hailuo (4.5x), Seedance (0.9x), Gemini Omni Flash (6.2x), Grok Imagine Video (3.3x), Kling Video (7x), Leonardo Motion (1.1x), Luma (8.3x), MiniMax (8.3x), Runway Video (4.2x), Veo (13.3x)
Video to Video: Seedance (0.9x), Gemini Omni Flash (6.2x), Grok Imagine Video (3.3x), Kling Video (7x), Runway Video (4.2x)
Other filters
Moderated models currently include Auto, the Claude models, GPT-5 models, GPT Image, o4 Mini, Nova Pro, and Perplexity Sonar Pro Search. Everything else is listed as Unmoderated.
You can also filter chat models by Actions, Vision, Agentic, and Reasoning. Voice GPT is Magai's voice model (32K context, 1x).