AutoRouter (@Auto_Router) header
AutoRouter (@Auto_Router) profile picture

Cut your AI bill in half. Use your existing subscriptions. Every leading model inside Codex and Claude Code, and the right one for every task. Beta for macOS.

Joined October 2026
Photo by AutoRouter (@Auto_Router)Photo by AutoRouter (@Auto_Router)Photo by AutoRouter (@Auto_Router)Photo by AutoRouter (@Auto_Router)Photo by AutoRouter (@Auto_Router)Photo by AutoRouter (@Auto_Router)
Global custom routing rules in ChatGPT Set which tasks need which models from the prompt bar. AutoRouter remembers them and routes across agents Plugin install here: chatgpt.com/plugins/plugin_a…
1
1
3
465
the routing department approves
we built a ChatGPT plugin! natively hook @Auto_Router into ChatGPT in one click, and you get: 🌈 all the models from all the providers 🤖 automatic routing that saves you money 💵 sub stacking that gets you more weekly usage try it, I dare ya! chatgpt.com/plugins/plugin_a…
2
77
AutoRouter retweeted
we built a ChatGPT plugin! natively hook @Auto_Router into ChatGPT in one click, and you get: 🌈 all the models from all the providers 🤖 automatic routing that saves you money 💵 sub stacking that gets you more weekly usage try it, I dare ya! chatgpt.com/plugins/plugin_a…
2
3
1
16
1,098
AutoRouter retweeted
install @Auto_Router. keep using your favorite harnesses. save up to 50% on your AI bill. download now: autorouter.com or DM to get AutoRouter for your team.
1
2
9
1,078
AutoRouter picks the model for each turn in Codex, Claude Code, Claude Desktop and T3 Code today, on the plans you already pay for.
the model dropdown will be dead by the end of the year
1
1
5
261
Replying to @pandez_
we live to serve 🫡
29
Replying to @AdamHoltererer
the syntax style in the generated code usually gives away the teacher model faster than any public benchmark score
1
99
you can now route to Crouton Alpha through AutoRouter. restart your harness and it shows up in the model picker for Claude Code and other clients
Crouton Alpha (stealth model) is free for the next week 1M Context Multi-modal No training on data Limited to certain regions
5
403
Replying to @zainhas
DeepSeek at 10 and GPT-6 Astra at 60 is why routing every prompt to the right model matters
19
Replying to @MatthewBerman
we built AutoRouter to automate this exact setup inside Claude Code and Claude Desktop
12
Replying to @MatthewBerman
forgetting to switch back from Opus to Haiku after a code review wastes your daily quota on grep commands
39
Replying to @opencode
1m context for free means my entire monorepo is going in the toaster
5
1,016
Replying to @trq212
not checking for bugs and just assuming it works is already how i ship code anyway
13
Replying to @levelsio
some enterprise sales rep just fell to his knees in a patagonia vest
31
Replying to @steipete
rip to the ephemeral void where i asked how to center a div
1
150
Replying to @rauchg
my favorite part of ci is not having to configure ci
76
Sub Stack is live. Go break a usage cap
today we're shipping Sub Stack for Claude and ChatGPT subscriptions it intelligently rotates between your subscriptions to keep your chat going if one maxes out giving you a unified quota for all your AI logins, on any harness
1
3
164