Models

Frontier models. One simple choice.

Tsedal AI brings models from eight leading AI labs into your editor. You don't manage models: choose a level, or leave it on Auto, and Tsedal picks the right one for each request.

Recommended

Auto

Tsedal picks the level for each request: Fast for everyday work, a quick model for small chores, and Smart when a question looks hard.

  • Most of your work
  • When you'd rather not choose
  • Mixing quick questions and hard ones
Speed
Depth

On every plan, Free included.

Model

Bars show what's left today

The labs

Built on eight frontier labs

Tsedal AI uses models from these labs and chooses among them for each request. As they release better models, Tsedal moves to them, with nothing for you to change.

  • ClaudeAnthropic
  • GPTOpenAI
  • GeminiGoogle
  • DeepSeekDeepSeek
  • QwenAlibaba
  • GLMZ.ai
  • KimiMoonshot AI
  • MiniMaxMiniMax
Levels

Four levels, for every kind of task

Choose a level in the message box. Every level works in Ask, Plan and Agent.

Default

Auto

Tsedal picks the level for each request.

  • Everyday work goes to Fast
  • Small chores to a quick model
  • Hard questions move up to Smart
Speed
Depth

Every plan

Everyday

Fast

Answers in a second or two.

  • Questions about your code
  • Edits and small agent tasks
  • Commit messages and names
Speed
Depth

Every plan, daily

Reasoning

Smart

Thinks the problem through first.

  • Tricky bugs and stack traces
  • Refactoring and bigger changes
  • Questions about screenshots
Speed
Depth

A taste on Free, daily on Pro and Max

Frontier

Max

The most capable models we offer.

  • The hardest problems
  • Long agent tasks across many files
  • Architecture and design
Speed
Depth

A taste on Free, daily on Pro, most on Max

How it works

How Tsedal picks a model

  1. 1You choose a levelOr leave it on Auto, and Tsedal chooses the level for each request.
  2. 2Tsedal picks the modelYour request goes to the model that suits it best on that level.
  3. 3A busy model never stops youIf one is busy or down, another takes over. A request that can't be answered isn't counted.

The chat in your editor always shows which model answered. If the shared monthly budget is nearly used up, Max requests may run on Smart for a while.

Always current

New models, without changing a thing

AI labs release better models every few weeks. Tsedal AI updates the models behind each level as they arrive, so you always work with current frontier AI. Your plan, your levels and your settings stay exactly the same.

New models on the blog
Frontier labs
8
Simple levels
4
Plan, paid in birr
1
Settings to change
0
Plans

Levels on each plan

LevelFreeProMaxBusiness
AutoIncludedIncludedIncludedIncluded
FastDailyGenerous dailyAbout 2× ProSized for your team
SmartA taste each monthDailyAbout 2× ProSized for your team
MaxA taste each monthDailySeveral times ProSized for your team
Agent tasksA few a day, on FastOn every levelOn every levelOn every level
Questions

About models

Which model answers my request?

Tsedal sends each request to the model that suits it best on the level you chose, from the labs it works with. The chat in your editor shows which model answered.

Can I choose a specific model?

You choose a level, not a model. That way Tsedal can move you to a better model as soon as one is released, and send your request elsewhere when a model is busy, without you changing anything.

Will the models change?

Yes. As labs release better models, Tsedal AI updates the models behind each level. Your plan, levels and allowances stay the same.

Do the labs keep my code?

Tsedal AI doesn't store your prompts, code or answers. Each request is processed by the model that answers it, under that lab's terms. Read our terms and privacy for the details.

Start building with Tsedal AI.

Start free. Move to Pro or Max, paid monthly in birr, when you need more.