Live role
Flagship reasoning
Astra handles complex instructions, decisions, and agent orchestration.
OpenAI models: GPT-6 Astra, GPT-Image-2, Sol, Terra, and Codex.
GPT-6 Astra is live for reasoning, GPT-Image-2 is live through the drawing Combo, and the configured Sol, Terra, and Codex profiles remain below the main picker.
Floatboat · OpenAI
Current model
OpenAI
What do you want to finish?
Reason through this decisionImplement and verify this featureTurn this brief into an image
Describe the outcome or add project files…
- Live chat model
- GPT-6 Astra
- Live image model
- GPT-Image-2 via Combo
- Configured capacity
- Sol · Terra · Codex
- API key
- Not required
OpenAI model characteristics
Astra is the OpenAI route for high-stakes reasoning in the client.
GPT-6 Astra is selectable for chat, GPT-Image-2 is available through the drawing Combo, and code-specialized capacity is configured without a main-picker entry.
Work style
Reasons before execution
Best where the cost of a weak plan is high.
Visual lane
Delivered through a Combo
GPT-Image-2 produces images rather than chat output.
Status discipline
Configured is not selectable
Sol, Terra, and Codex remain below the main picker.
Role-based lineup
Do not treat every GPT model as interchangeable.
The useful distinction is the job each model is meant to do, not only its generation number.
gpt-6-astra
Strong reasoning and broad instruction following.
Client picker
gpt-5.6-sol / gpt-5.6-terra
1M context and 272K cliff profiles configured below the picker.
Not in the main picker
gpt-5.3-codex / gpt-5.3-codex-spark
Code-specialized models with 400K / 128K capacity configuration.
Not in the main picker
gpt-image-2
High-quality image, illustration, and cover generation.
Live · ai-image-drawing
Outcome-based routes
Use the live OpenAI capability that matches the deliverable.
A reasoning task and an image brief should not be forced through the same product entry.
Complex product decision
Task input
Resolve competing requirements before committing to implementation.
Processing focus
Astra follows constraints, compares consequences, and builds an actionable direction.
Expected result: A reasoned decision ready for the project team.
Agent orchestration
Task input
Coordinate a multi-step task that mixes planning and execution.
Processing focus
Astra keeps the instruction hierarchy in view while the task progresses.
Expected result: A coherent run instead of disconnected answers.
Image from a creative brief
Task input
Produce an illustration, cover, or campaign visual.
Processing focus
The `ai-image-drawing` Combo sends the visual task to GPT-Image-2.
Expected result: A visual asset rather than another text response.
OpenAI in Floatboat
One provider family, accurately separated by product surface.
Chat and image generation use different entry points. Configured coding models are not currently selectable in the main client picker.
Frontier reasoning
GPT-6 Astra is selectable for complex reasoning and broad instruction following.
Agent task execution
Astra is positioned for deep Agent orchestration and architecture-level work.
Image generation
GPT-Image-2 is live as the default high-quality model in `ai-image-drawing`.
Prepared code capacity
Sol, Terra, and Codex profiles are configured but not exposed in the picker.
Choose the available route, not only the model name.
Choose GPT-6 Astra when
- The task needs strong reasoning before implementation.
- You are coordinating several dependent Agent steps.
- Instruction following and architectural judgment matter together.
Use another Floatboat route when
- The deliverable is an image; open the drawing Combo.
- You need faster routine processing; consider a Flash model.
- Do not select Sol, Terra, or Codex from the main picker; they are not exposed there.
Choose by outcome
Reason, build, respond, or generate.
Astra
Use when the task needs careful reasoning before action.
Sol
Use for implementation-heavy coding and sustained technical work.
Codex
Configured code-specialized capacity below the main picker.
GPT Image 2
Use when the deliverable is visual rather than another block of text.
Explore Other Model Families
Compare model families by the kind of work they support. Availability differs by model and entry point.
OpenAI · FAQ
Which OpenAI model paths are available or configured in Floatboat?+
The matrix covers GPT-6 Astra, GPT-Image-2, GPT-5.6 Sol and Terra, and GPT-5.3 Codex and Codex Spark, with their distinct roles and status.
What work is GPT-6 Astra intended for?+
Astra is the live OpenAI option for flagship reasoning, complex instruction following, architecture decisions, and deep Agent orchestration.
Why is GPT Image 2 listed with text models?+
It represents a different output lane—image generation—within the same broader workflow and project context.
What can GPT-Image-2 generate?+
GPT-Image-2 is positioned for high-quality image generation, illustrations, cover artwork, and other visual deliverables.
What are GPT-5.3 Codex and Codex Spark for?+
They are code-specialized evaluation profiles with configured 400K and 128K capacity, respectively.
Pick Astra for reasoning or GPT-Image-2 for visuals
Use GPT-6 Astra in chat for reasoning and GPT-Image-2 through the image Combo for visuals; other variants are not default picker choices.