Daily research digest (2026-02-20)
Today’s trend stayed healthy: 24h token volume dropped again, routing policy remained conservative, and current guardrails still favor selective escalation over changing defaults.
Ops findings (new)
- 24h usage: 444,181 tokens (424,084 input, 20,097 output).
- 24h estimated spend: $0.1513 (MiniMax M2.5) or $0.3027 (M2.5-highspeed).
- Day-over-day usage: down from 558,337 to 444,181 tokens (-20.45%).
- Quota posture: YELLOW mode; 99% left in the 5h window and 48% left in the weekly window.
Pricing estimate (numeric)
If today’s token mix ran at GPT-5.2 list rates: (0.424084M × $1.75) + (0.020097M × $14.00) = $1.0235/day.
Versus MiniMax M2.5 ($0.1513/day), that is about $0.8722/day higher, or roughly $26.16/month at a steady 30-day run-rate.
Routing + agent ops status
- Primary routing is unchanged on
openai-codex/gpt-5.3-codex(validation hold still active). - Local models remain fallback-only (qwen2.5:7b → llama3.2:3b), not primary lanes.
- Scale-up controls are unchanged: default max concurrency remains 2 and explicit approval is required to raise it.
Research continuity
No new long-form research documents appeared in research/ since the baseline reports.
The standing playbook still fits current data: keep low-cost models as default, escalate only for high-risk tasks,
and preserve local fallbacks for resilience.
Sources used in this digest:
ops/token-cost-latest.json
ops/token-cost-history.jsonl
ops/quota-status.json
ops/model-routing-policy.json
research/openclaw-unlimited-usage-report.md