Nasiko Open Runtime for Agents Cuts Tool-Output Tokens 42% With No Accuracy Loss on Logs
aakashgupta · x · 2026-11-17
Nasiko launched an open runtime platform for agents (11.9K GitHub stars), bundling TokenOps, gateway, budgeting, and runtime controls in one deploy:
- Token compression: compresses tool outputs (logs, search results, records) before the model reads them. Across 50 questions on real system logs, input tokens dropped from 617,202 to 357,958 (42% fewer), and cost per correct answer fell from $0.037 to $0.022 (40% lower) with accuracy unchanged (44/50).
- Trade-off on records: on 102 record lookups, correctness slipped from 100 to 96 after compression.
- Also included: an Agent Gateway for multi-model access, budgets with alerts to stop overruns, and Runtime Controls to allow/ask/block each tool call.
Installable via a one-line CLI command with observability and cost metering wired in.
Related event: Nasiko Launches Open-Source AI Agent Runtime Platform with TokenOps Focus(3 posts)→
More from coding & agent
- Open-source Pascal Editor brings near-native 3D building design to the browser via WebGPU — anthara_ai · 2026-11-17
- Prompt of the day: have your agent study all skills and build a skill orchestrator — yungcontent · 2026-11-17
- OpenClaw cut CI runner-hours by 86.5% and compute value by 95.5%: here's how — vincent_koc · 2026-11-17
- Databricks benchmark: Genie passes 91% of enterprise data tasks vs 45–58% for frontier agents — matei_zaharia · 2026-11-17
- Stock Screener API launches on x402: agents screen 17,000 stocks for $0.02 — virattt · 2026-11-17
- Dev calls for an 'omni-feed' product: one filterable pane for all agent actions — DavidWells · 2026-11-17