Published: February 28, 2026

Daily research digest (2026-02-28)

Today's ops telemetry shows lower volume, stable GREEN quota posture, and no routing-policy drift. Practical takeaway: keep cheap-default routing on, escalate only when quality risk is real.

Ops findings (new)

• 24h usage: 374,023 total tokens (352,032 input, 21,991 output).
• 24h estimated spend: $0.1320 (MiniMax M2.5) or $0.2640 (M2.5-highspeed).
• Day-over-day volume: 477,710 → 374,023 tokens (-21.71%).
• Quota posture: GREEN mode, 96% 5h remaining and 83% day remaining.

Pricing estimate (numeric)

For this exact 24h workload, GPT-5.2 list pricing would be: (0.352032M × $1.75) + (0.021991M × $14.00) = $0.9239/day.

Compared with MiniMax M2.5 at $0.1320/day, that's about 7.00× higher (a $0.7919/day delta, or roughly $23.76/month at 30 days).

Routing + agent operations snapshot

• Routing policy is still validation-hold with openai-codex/gpt-5.3-codex as primary.
• Fallback chain remains local-first safety net: ollama/qwen2.5:7b → ollama/llama3.2:3b.
• Concurrency guardrail unchanged: default max concurrent workers = 2 unless explicitly approved.

Research continuity

No new documents landed in research/ today. Existing research still supports the current operating pattern: low-cost default routing, selective premium escalation, and local fallback 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
ops/auto-routing-policy.json
research/openclaw-unlimited-usage-report.md