TLDRocket
Sign in

Weave Router 2.0

Product Hunt Ben Lang

Weave Router 2.0 sends coding agent jobs to the cheapest model with room left. It says it matches GPT-6 Astra on two benchmarks for half the cost and twice the speed.

Based on reporting by Product Hunt, Ben Lang — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Weave Engineering Intelligence has launched Weave Router 2.0, its third release, and the pitch is simple: stop wasting model calls. The system routes each coding-agent request to the cheapest model that can still get the job done, while also respecting whatever quota is left on the user’s existing plans.

That means Claude models can be used in Codex, and GPT models can be used in Claude Code. Weave says the router is “subscription aware,” which is a neat way of saying it tries to make the most of what people already pay for instead of forcing them into one stack.

The company is backing the claim with benchmark results. On Terminal-Bench 4.0 and SWE-Atlas, it says Weave Router 2.0 matches GPT-6 Astra at half the cost and runs 2x faster. Those are strong numbers if they hold up in real use, and they point to a very specific kind of efficiency: not just cheaper inference, but smarter switching.

That switching is driven by a new classifier that scores task complexity, plus what Weave calls ache-aware switching, where the router only changes models when the savings beat the rebuild cost. That last part matters. A lot of “smart routing” falls apart when the handoff itself becomes the expensive bit. Weave is trying to keep that from happening.

My take — AI-written commentary, not fact-checked reporting

This is the sort of routing trick that actually deserves attention because it attacks the bill, not the brochure. The industry loves to sell bigger models and call it progress; meanwhile, the grown-up move is often knowing when not to use them. The real winner here is boring efficiency, which is usually where the money is hiding.

Read more about this at: Product Hunt

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.