@0xAnto Exactly the failure mode. In OpenFusion a judge reviews the cheap draft and escalates only when needed, and failing tests or tool loops bump the tier automatically, so you don't pay twice for everything. Open source, works in Claude Code/Codex: calesthio/OpenFusion
11 views
@ibuildthecloud Open weights will drive cost down, but routing still matters: why spend Opus on a grep? That's why I made it harness-agnostic and open source instead of a Copilot feature. It routes to whatever models you have, open-weight included.
calesthio/OpenFusion
9 views
@kimmonismus Agree. And it doesn't have to stay inside Copilot. I built an open-source version that gives Claude Code, Codex, Cursor and Gemini the same per-task model routing with a judge that escalates.
calesthio/OpenFusion
27 views
One command: uvx openfusion install claude-code . (swap in codex, cursor, github-copilot, gemini-cli, antigravity, opencode).
Would love benchmarks on other harnesses. A ⭐ helps a lot:
calesthio/OpenFusion
48 views
OpenFusion: Efficient sub task routing
preview.redd.it
The main agent routes each subtask to the cheapest model that can do it well, and has a judge escalate to a stronger model when the cheap attempt isn't good enough. Would also really appreciate if someone could benchmark using some other harnesses.
1 point
OpenFusion for sub task routing
Using this your main agent routes each subtask to the cheapest model that can do it well, and has a judge escalate to a stronger model when the cheap attempt isn't good enough.
Open Sourcing this. You can use it if it helps you with token and time saving. I am seeing about 25% cost and 25-40% time saving for complex multi-step tasks within Codex
Would also really appreciate if someone could benchmark using some other harnesses.
calesthio/OpenFusion
reddit.com
2 comments
Introducing OpenFusion for appropriate model routing
It routes each subtask to the cheapest model that can do it well, and has a judge escalate to a stronger model when the cheap attempt isn't good enough.
Open Sourcing this. You can use it if it helps you with token and time saving. I am seeing about 30% token and 25-40% time saving for complex multi-step tasks within Claude Code.
Would also really appreciate if someone could benchmark using some other harnesses.
calesthio/OpenFusion
1 point · 3 comments
Mentions
@0xAnto Exactly the failure mode. In OpenFusion a judge reviews the cheap draft and escalates only when needed, and failing tests or tool loops bump the tier automatically, so you don't pay twice for everything. Open source, works in Claude Code/Codex: calesthio/OpenFusion
@ibuildthecloud Open weights will drive cost down, but routing still matters: why spend Opus on a grep? That's why I made it harness-agnostic and open source instead of a Copilot feature. It routes to whatever models you have, open-weight included. calesthio/OpenFusion
@kimmonismus Agree. And it doesn't have to stay inside Copilot. I built an open-source version that gives Claude Code, Codex, Cursor and Gemini the same per-task model routing with a judge that escalates. calesthio/OpenFusion
One command: uvx openfusion install claude-code . (swap in codex, cursor, github-copilot, gemini-cli, antigravity, opencode). Would love benchmarks on other harnesses. A ⭐ helps a lot: calesthio/OpenFusion
OpenFusion: Efficient sub task routing preview.redd.it The main agent routes each subtask to the cheapest model that can do it well, and has a judge escalate to a stronger model when the cheap attempt isn't good enough. Would also really appreciate if someone could benchmark using some other harnesses.
OpenFusion for sub task routing Using this your main agent routes each subtask to the cheapest model that can do it well, and has a judge escalate to a stronger model when the cheap attempt isn't good enough. Open Sourcing this. You can use it if it helps you with token and time saving. I am seeing about 25% cost and 25-40% time saving for complex multi-step tasks within Codex Would also really appreciate if someone could benchmark using some other harnesses. calesthio/OpenFusion reddit.com
Introducing OpenFusion for appropriate model routing It routes each subtask to the cheapest model that can do it well, and has a judge escalate to a stronger model when the cheap attempt isn't good enough. Open Sourcing this. You can use it if it helps you with token and time saving. I am seeing about 30% token and 25-40% time saving for complex multi-step tasks within Claude Code. Would also really appreciate if someone could benchmark using some other harnesses. calesthio/OpenFusion