News July 9, 2026
Introducing Wave-1 Thinking
Wave-1 can now reason before it answers. Three levels let you pick how much thought a question deserves, and every step of the reasoning streams where you can read it.
Some questions want an answer right away. Others deserve real thought: a tricky bug, a proof, a plan with edge cases. Until today, Wave-1 treated both the same way. Starting now, the chat has a reasoning control, and you decide.
Three levels
- Flash is the Wave-1 you already know. It answers right away and stays the default, because most questions are Flash questions.
- Thinking has the model reason through the problem first, then answer. It is noticeably stronger on math, code, and anything with steps.
- Ultra is the deepest level. Wave-1 takes its time, considers alternatives, checks its own work, and only then writes the answer.
Switching takes one tap on the level chip next to the message box. Your choice sticks until you change it, in English and Arabic alike.
Reasoning you can read
When Wave-1 thinks, you see it thinking. The reasoning streams live in a panel above the answer, then folds away into a small note that tells you how long it took. Tap it any time to read the full chain of thought.
A better place for code
Alongside thinking, the chat became a better place to write code. Code blocks are now syntax highlighted in the languages you actually use, labeled with their language, and copied in one click. Paired with Thinking or Ultra, Wave-1 plans the code before it writes it.
Safety does not change with the level
Every message still passes through Wave Guard before any reasoning starts. The safety bar is identical at Flash and at Ultra: thinking harder never means thinking around the rules. If you find something that got through, tell us at wave@waveai.cloud.
