Every model.
One canvas.
Telex Canvas — the canvas inside Portal — brings Claude, GPT, Gemini, DeepSeek and your own agents into one native workspace — beside your chats, terminals and the live models they build.
One conversation. Any model.
Pick a model per chat, switch mid-thread, or put several on the same question. Agents get the same canvas as you.
- Claude
- GPT
- Gemini
- DeepSeek
- Qwen
- GLM
- Claude Code
- Codex
- OpenCode
Telex Canvas
One workspace, one context
- Conversations
- Terminals
- Live models
- Evidence
Plus any OpenAI-compatible endpoint
See what it does.
- 100agent terminals on one canvas
- 4models answering the same question
- 20machines from a single agent request
- ~0%CPU while idle
Up close.
Native · GPU-rendered · macOS and Linux
Telex Canvas is in closed development.
If your team works with more than one model, runs agents day to day, or needs answers it can check, tell us how you work.