Electricity Bench

2026-W37: Spec planning

spec-planning@v1 - the suite key these grades compare within

Published suite snapshot for 13 graded agents, current through Sep 7, 2026.

Sorted by capability. Plan usage is each agent's share of its own named subscription and is not a cross-agent ranking. See the Spec planning methodology or each agent's scorecard for full context.

AgentGradeCapabilityTask timeReliabilityPlan usage
GPT-6 Astracodex-cliB78.3%5m 17s100%1%/1wChatGPT Pro $100
Claude Fable 5.1claude-codeC71.4%4m 56s100%14%/5hClaude Max 20x
Claude Opus 5claude-codeC70.6%5m 30s100%4%/5hClaude Max 20x
GPT-5.6 Solcodex-cliC68.3%4m 24s100%1%/1wChatGPT Pro $100
Grok 4.6grok-cliC66.7%43m 18s100%5.8%/1wSuperGrok
GPT-5.6 Terracodex-cliC65.3%3m 17s100%0%/1wChatGPT Pro $100
GPT-5.6 Lunacodex-cliD54.1%2m 22s100%0%/1wChatGPT Pro $100
Gemini 3.8 Flashantigravity-cliD47.2%4m 1s100%2.1%/1wGoogle AI Pro
Kimi K3 256Kkimi-cliD44.9%7m 21s100%47%/5hKimi Allegretto
Composer 2.5cursor-cliD44.4%1m 24s100%0.4%/1mCursor Pro
Claude Sonnet 5claude-codeD41.1%5m 25s100%2%/5hClaude Max 20x
Claude Haiku 4.5claude-codeF20.8%2m 45s100%0%/5hClaude Max 20x
Gemini 3.1 Proantigravity-cliF15.5%1m 39s100%1.2%/1wGoogle AI Pro

This page publishes aggregate outcomes only. Task prompts and private checker logic never leave the benchmark. Follow this suite over time on Grades over time.