LunaRoute sells concurrency, not tokens: fixed-price inference for coding agents on open models
dbreunig · x · 2026-11-08
Blogger dbreunig strongly recommends LunaRoute, an inference provider that bills by concurrent lanes rather than tokens, offering fixed monthly-price inference on supported open models so agent iterations don't rack up token charges.
Key points:
- His Opus 5.5 usage is a small share; most tokens go to less complicated pipeline work on cheap open models (ds4-flash, glm 5.3) and a tweaked Hermes agent
- LunaRoute: zero retention on standard prompts/responses, never used for training; from $99/mo with no token cap (queue limits apply), external providers billed separately
- Pitch: let coding agents 'build until it works' without metering every retry
Related event: Blogger Mixes Open-Source Models to Cut Token Costs(2 posts)→
More from coding & agent
- Kinference routing benchmark: cache-aware routing cut TTFT from 6.88s to 0.31s on shared prefixes — shoumikgoswami · 2026-11-08
- Claude wrote a wrong Pokémon landmark note and kept trusting it — Anthropic tested a memory repair pass — shoumikgoswami · 2026-11-08
- Making Interactive HTML Pages With AI Instead of Boring PDFs — jonathanmast · 2026-11-08
- Orchestrating subagents with Opus 3: models love it, but Claude Code overhead kills it — repligate · 2026-11-08
- Agents audit a dentist quote in 20 seconds, likely saving $1k — gabriberton · 2026-11-08
- Yacine: if your merge takes weeks, you're not working on anything important — yacineMTB · 2026-11-08