Dev benchmarks Astra vs Opus 5.5 workflows: combo tops solo at 82.8/100 but costs more
kevinkern · x · 2026-09-27
Developer Kevin Kern ran the same feature in his finance app through solo runs and a dozen Astra/Opus/Luna orchestration combos, scoring each on code (50%), UX (25%) and design (25%):
- Best overall: Astra high → Opus 5.5 → Astra high review, 82.8/100 (84 code, tied best)
- Best value: Opus 5.5 + Astra high review, $32.48 est. API cost
- Best design: Astra high → Luna → Opus 5.5, 86/100 design
- Luna was cheapest but scored lowest on code quality; solo Opus/Astra runs landed close to combos at much lower cost
He cautions it's a small personal experiment and results may differ with many parallel threads. Screenshots and code for each run are linked.
Related event: Hands-on test: Astra orchestrating Opus 5.5 scores best but doubles cost(8 posts)→
More from coding & agent
- Personal agents from big labs vs one shared company agent: which design wins? — AshishKulkarni1411 · 2026-11-16
- 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