Tokenwise
A smart LLM proxy that shows where you're overpaying
186 followers
A smart LLM proxy that shows where you're overpaying
186 followers
Tokenwise is a one-line LLM proxy (OpenAI-compatible baseURL) for makers and small teams. It learns from your real requests, shows exactly where you're overpaying, proven with quality checks on your own traffic, not public benchmark, and lets you apply the fix in one click while it verifies the savings in real dollars.
Mailwarm
Does Tokenwise break down coding agent spend by session or only by model?
Tokenwise
Hey @bengeekly not by session yet, but the workaround works well today.
You can pass a tag (or session ID) on each request via the X-Tokenwise-Tags header, and Tokenwise clusters all requests sharing that tag, so for a coding agent you'd set X-Tokenwise-Tags: session-{conversationId} and see the full cost breakdown per session: total spend, which model, which prompt template (we cluster those semantically too), outliers, etc.
First-class sessions view (auto-grouped, no header needed) is on the short roadmap.
Great idea, I've had the same problem and spent many hours debugging transcripts to look for token savings.
Tokenwise
@willsmithte Ha yeah, that's literally why this exists. Digging through transcripts by hand to find the one expensive prompt is the worst. Tokenwise just shows you cost per prompt so you skip the manual hunting. Appreciate it.
The "quality check on your own traffic" angle is genuinely the right frame. I've tried a few LLM cost tools and most just show you aggregate spend with generic benchmark comparisons — which is basically useless when your prompts are domain-specific.
One thing I'd love to see: for coding agents specifically, the spend is often bursty and session-based. A 2-hour Claude Code session can easily hit $5-10, but you don't know *which* part of the session burned the tokens. Was it the initial codebase ingestion? A long loop of test-fix retries? Breaking that down by sub-task within a session would be way more actionable than just "session-xyz cost $8.40."
Tokenwise
@justin_winter Just checked and it's loading fine on my end, and fast too. Might've been a quick blip, mind trying again or a hard refresh? If it's still down for you let me know what you're seeing and roughly where you are and I'll dig into it.
honestly never dealt with LLM billing chaos myself but i can see why this exists, the amount of devs who complain about surprise API bills is insane
the fact that it tells you what to cut instead of just showing you a graph is the actual useful part. dashboards are useless if they don't tell you what to do next
and checking quality before cutting costs is something no one else is doing most tools just say 'use gpt-3.5' and hope for the best.