Flash, Max and Heavy
The three tiers (efforts), how they differ and which to pick.
- Are Flash, Max and Heavy different models?No. They share SuperCosmos 1.1's weights and 1M-token window; each tier has its own tuned instructions and thinking budget.FAQ
- Can I see what Max or Heavy is thinking?On the Model API, yes: reasoning comes back as reasoning_content. It's billed as output.FAQ
- Can I switch tiers in the middle of a conversation?Yes. Line CLI switches with /effort; on the Model API, send the next request with a different model id and the same history.FAQ
- Does Flash think before answering?Flash has no extended thinking: it answers straight away. Max and Heavy reason first and return reasoning_content.FAQ
- Does Heavy know more facts than Flash?No. All tiers share the same weights, so they know the same things. Heavy reasons longer about them.FAQ
- How do I know when to move up a tier?Move up when a task keeps failing its own checks, needs a plan across many files, or answers miss the point.FAQ
- How much slower is Heavy than Flash?In Benchmark 2, median time per task was 8.7 s on Heavy, 7.2 s on Max and 3.5 s on Flash.FAQ
- Is Heavy always better than Flash?No. Heavy thinks longer, which helps on hard problems, but on everyday tasks Flash is faster, cheaper and often just as good.FAQ
- What does "auto" do?auto lets the product choose the tier per step. In Line CLI it routes between Flash and Max; Agent API auto runs bill as Flash.Explainer
- What is an "effort" (or tier)?An effort is how much work SuperCosmos puts into each step: how long it thinks, how carefully it plans and checks.Explainer
- What is SuperCosmos Flash, and when should I use it?Flash is the fast tier: no extended thinking, $1.60 / $3.40 per 1M tokens. Use it for quick edits, answers and fast agent loops.Explainer
- What is SuperCosmos Heavy, and when should I use it?Heavy is the deliberate tier: it thinks longest, $2.00 / $4.40 per 1M tokens. Use it for the hardest problems only.Explainer
- What is SuperCosmos Max, and when should I use it?Max is the strong tier: it thinks before acting, $1.80 / $3.80 per 1M tokens. Use it for planning, debugging and multi-file changes.Explainer
- What's the difference between Flash, Max and Heavy?Same model and 1M window; Flash answers without extended thinking, Max thinks before acting, Heavy thinks longest. Prices differ.Explainer
- Which SuperCosmos tier is used if I don't choose?Agent API: auto (billed as Flash). Model API family id: Flash. Line CLI: your saved effort, or auto.Reference
- Which tier is best for debugging?For debugging from logs and failing tests, start with Max. Here's why, and when to switch.FAQ
- Which tier is best for small code edits?For small, well-described code changes, start with Flash. Here's why, and when to switch.FAQ
- Which tier should I use for a big migration?For large builds and migrations, start with Heavy. Here's why, and when to switch.FAQ
- Which tier should I use for a code or design review?For architecture and design reviews, start with Heavy. Here's why, and when to switch.FAQ
- Which tier should I use for a high-volume app?For high-volume apps (classification, short replies, chatbots), start with Flash. Here's why, and when to switch.FAQ
- Which tier should I use for an agent loop?For agent loops with many short steps, start with Flash. Here's why, and when to switch.FAQ
- Which tier should I use for everyday questions?For everyday questions and chat, start with Flash. Here's why, and when to switch.FAQ
- Which tier should I use to ask questions about my code?For questions about a codebase, start with Flash. Here's why, and when to switch.FAQ
- Which tier should I use to build a new feature?For building a feature end to end, start with Max. Here's why, and when to switch.FAQ
- Which tier should I use to set up CI or deployments?For setting up projects, CI and deployments, start with Max. Here's why, and when to switch.FAQ
- Which tier should I use to write a script?For scripts and one-off automation, start with Flash. Here's why, and when to switch.FAQ
- Which tier should I use when a mistake would be costly?For tasks where a wrong answer is expensive, start with Heavy. Here's why, and when to switch.FAQ
- Why are the tiers also called Fast, Strong and Deliberate?They're plain-language labels: Flash = Fast, Max = Strong, Heavy = Deliberate.FAQ
Still need help?
Tried the steps above and still stuck? Email us. We usually reply within the first hour, and always within 24 hours. A full support team is with you.
Infrastructure questions, or permanent deletion of your data
infrastructure@clatterlabs.xyzFor deleting your whole account or the data of one service.