Pairing Astra and Opus for coding: 3-stage pipeline tops at 82.8, but solo runs cost far less
kevinkern · x · 2026-09-27
The author tested Astra, Opus 5.5, and Luna — solo and in various orchestration pipelines — building the same feature in his finance app, scoring 50% code quality / 25% UX / 25% design.
Key results
- Best overall: Astra high → Opus 5.5 → Astra high review, 82.8/100
- Best value: Opus 5.5 building + Astra high review, $32.48 estimated API cost, 80.5
- Best design: Astra high → Luna → Opus 5.5, 86
- Best code: Opus 5.5 + Astra high review, 84 (tied with the 3-stage pipeline)
Takeaways
- Luna was by far the cheapest but scored lowest on code quality.
- Astra and Opus solo runs landed close to combined runs at much lower cost.
- Multi-model orchestration isn't always worth it: the review-only combo is the value pick, and only the full 3-stage pipeline edged it out.
He published a dozen concrete orchestrations (Astra delegating to Opus, Opus on backend, Astra on frontend + review, etc.) for reproduction.
Related event: Hands-on test: Astra orchestrating Opus 5.5 scores best but doubles cost(8 posts)→
More from coding & agent
- WebArena-Pro launches: 300-task multimodal web-agent benchmark across 20 apps — xhluca · 2026-11-16
- Self-improving agent papers this week: agents learn the benchmark, not the task — TheTuringPost · 2026-11-16
- 1,000 agents, one week, 630k lines: HF builds a weights-free Python chatbot — _lewtun · 2026-11-16
- One-person content agency targets $30K/mo with 11 AI agents — full playbook — VibeMarketer_ · 2026-11-16
- Hugging Face releases definitive fine-tuning guide for 3D generation model TRELLIS.2 — ariG23498 · 2026-11-16
- Continual-ARC: lifelong ARC-AGI stream finds agent harness persistence rarely beats simple caching — unifyai · 2026-11-16