Change the model without rebuilding the workspace.
Ready out of the box with no API key required. Use Auto Mode, available to everyone, or directly select models enabled by your subscription configuration.
Available paths
Models available in the client
Use Auto routing or directly select frontier models from Anthropic, OpenAI, Google, and DeepSeek.
Auto Mode
Let Floatboat route each task to a suitable model and fall back when needed.
Claude
Opus for deep architecture and Sonnet for high-performance everyday agent work.
claude-opus-4-6 · claude-sonnet-4-6 · claude-opus-5 / claude-sonnet-5 / claude-fable-5
OpenAI
GPT-6 Astra is live in chat; GPT-Image-2 is live in the drawing Combo; coding models are configured below the picker.
gpt-6-astra · gpt-5.6-sol / gpt-5.6-terra · gpt-5.3-codex / gpt-5.3-codex-spark · gpt-image-2
Gemini
Pro for long-context multimodal analysis; Flash for fast, high-throughput work.
gemini-3.1-pro-preview · gemini-3.8-flash · gemini-3-pro-image-preview · gemini-2.5-flash-image / gemini-3.1-flash-image-preview
DeepSeek
V4 Pro for complex code and mathematics; Flash for efficient everyday execution.
deepseek-v4-pro · deepseek-flash
Multimedia Combos
Live image and video engines
These models run through built-in Combo Skills rather than the main chat-model dropdown.
gpt-image-2
High-quality image, illustration, and cover generation.
OpenAI
gemini-3-pro-image-preview
Complex generation, precision editing, multi-image redraw, and style fusion.
Gemini
gemini-2.5-flash-image / gemini-3.1-flash-image-preview
Fast image generation and local edits.
Gemini
doubao-seedance-2-0-260128
Text-to-video, image-to-video, frame control, and synchronized audio.
Seedance
doubao-seedance-2-0-fast-260128
Fast drafts and storyboard clips.
Seedance
Configured capacity
Configured below the client picker
These models have capacity budgets or protocol adapters in place, but are not exposed in the main client picker.
Kimi
Kimi K3 has a configured 1M long-text reasoning profile below the picker.
kimi-k3
Qwen
Qwen 3.7 Max has a configured 1M long-text capacity profile below the client picker.
qwen-3.7-max
GLM
GLM 5 and GLM 4 support thinking-mode playback below the picker.
glm-5 · glm-4
MiniMax
MiniMax M2 has a configured 1M low-cost profile below the picker.
minimax-m2
Compare models by task and access path
See which models are selectable in chat, available through image or video Combos, or only configured below the picker before starting work.