Choose a model
Use the model control beneath the composer on desktop, web, or mobile. Search by name, filter by provider, or pin models you use often. Higher reasoning effort can improve results for complex tasks, but it takes longer and uses more credits. Start with the default, then increase when the task needs deeper planning or analysis. Ultra uses subagents to split larger work across parallel agents. Most tasks do not need Ultra.Recommended starting points
If you are unsure, start with Terra or Auto (when available on your plan). Move to Sol, Opus, or Max reasoning when quality matters more than speed.
Reasoning, Max, and Ultra
- Light / Low — quick, well-scoped tasks
- Medium — default balance of speed and depth
- High / Extra High — multi-step work with tradeoffs
- Max — give one model more time on the hardest single task (enable in settings if hidden)
- Ultra — parallel subagents when the work splits cleanly
Providers
Browse by lab. Each card jumps to that provider’s model list.
OpenAI
GPT 5.6 Sol / Terra / Luna, GPT-5.x, o3, o4-mini, Codex
Anthropic
Claude Opus 5, Sonnet 5, Opus 4.x, Haiku 4.5
Gemini 3.1 Pro, 3.5 Flash, 2.5 Pro / Flash / Lite
xAI
Grok 4.5, 4.3, 4.2, Build, Code Fast
Qwen
Qwen 3.7 Max / Plus, Coder, character models
Kimi
Kimi K3, K2.6, K2.7 Code
DeepSeek
DeepSeek V4 Pro and Flash
Z.ai
GLM 5.2, 5.1, 5, 4.7, 4.6
MiniMax
MiniMax M3, M2.7, M2.5
Cohere
Command A+, Command A, Reasoning, R7B
Perplexity
Sonar and Sonar Pro
Venice
Uncensored and roleplay-oriented models
Mistral
Mistral Small
Thesys
C1-enhanced Claude and GPT variants
Meta
Muse Spark 1.1
Inception
Mercury 2
Gemma
Gemma 4 31B
MiMo
MiMo V2.5 Pro and V2.5
Fireworks
GPT-OSS 120B and 20B
Together AI
Inkling
Model lists by provider
OpenAI
Flagship GPT family for coding, agents, multimodal chat, and reasoning.
Anthropic
Claude models for long context, careful writing, and agentic coding.
Gemini models with large context windows and strong multimodal support.
xAI
Grok models for fast frontier chat, coding, and agent work.
Qwen
Alibaba Qwen family for general chat, coding, and character workflows.
Kimi
Moonshot Kimi models for long-context coding and agent workflows.
DeepSeek
Z.ai
GLM models from Z.ai.
MiniMax
Cohere
Perplexity
Search-oriented Sonar models.
Venice
Uncensored and roleplay-oriented hosted models.
Mistral
Thesys
C1-enhanced variants for richer structured UI responses.
Meta
Inception
Gemma
MiMo
Fireworks
Together AI
Studio models
Image, video, and audio models live in Studio selectors (separate from the chat picker).Image
GPT Image 2 / 1.5 / 1 / Mini, Imagen 4, Nano Banana, Flux, Qwen Image, Grok Imagine, Seedream, Ideogram, Z-Image Turbo
Video
Seedance, Wan 2.6 Flash, Kling VIDEO 3.0, Grok Imagine Video
Music
Gemini Lyria
Speech and sound
ElevenLabs v3, Flash v2.5, and sound effects
Bring your own subscription (desktop)
On the ZeroTwo desktop app you can connect a provider subscription or API key and pick those models alongside the ZeroTwo catalog:
Connect providers in Settings → Desktop providers. Subscription-backed models use your connected account instead of the ZeroTwo credit catalog path.
Defaults and availability
- Availability depends on your plan. Free includes a limited set; paid plans unlock broader catalog access and monthly credits.
- Set a default model in app settings, or pin favorites in the picker.
- Work / Cowork and ZeroCode use the same catalog with work-agent capability filters — some models are chat-only.
- Exact IDs and metadata also appear on the public Models hub.
Related
- Pricing — plans, credits, and Auto allowance
- Prompting — get better results from any model
- Permission modes — control what agents can do
- Image generation — Studio image models
- Models hub — full public catalog