Bug Hunt Bench: GPT-6.2 Astra tops at 62.7, GPT-6.2 Sol matches Fable 5.2 at 6% cost
PawelHuryn · x · 2026-11-18
Pawel Huryn shares fresh Bug Hunt Bench (planted-bug fixing, verified blind) score-vs-cost data: GPT-6.2 Sol · max hits 54.7 for $7.00 (up +10.4 over 6.1), and Sol · xhigh ties Fable 5.2 · xhigh at 49 for 6% of the cost. GPT-6.2 Astra · max leads overall at 62.7. He argues OpenAI deserves more credit for per-iteration token efficiency, and describes his workflow: delegate implementation and review to Sol, use Opus 5.5 for planning, coordination, and judgement.
More from coding & agent
- LangChain's Deep Agents hits 50M downloads a year after betting on open agent harness — LangChain · 2026-11-19
- Running 15 AI Agents All Day Is Context-Switching Fatigue, IBM Engineer Explains — TejasKumar_ · 2026-11-19
- Dev's AI-native macOS dock with agent-managed modes flops; pivoting to to-do approach — trq212 · 2026-11-19
- Open-source browserd MCP server: 39% faster, 68% cheaper than Playwright MCP in real tests — Delicious_Break5937 · 2026-11-19
- Daiku Gives Your AI a Logged-In Browser So Multiple Agents Can Work Your Tabs in Parallel — TejasKumar_ · 2026-11-19
- Teams see LLM bills jump 2-3x after deploys — how do you catch cost regressions? — Junior-Guitar-290 · 2026-11-19