Opus 5's FrontierCode Score Drops from 53.4% to 48% in New Card
A developer flagged that Opus 5's reported FrontierCode score fell from 53.4% at launch to 48% on the new Opus 5.5 card; Andrew Carr explained the two runs used different thinking settings and the benchmark penalizes extra code changes.
2026-09-23 ~ 2026-09-23 · 3 related posts
- Episode 1: Claude Opus 5.5 Spotted in Claude Code Ahead of Imminent Launch(2026-09-23, 8 posts)
- Episode 2: Anthropic Launches Claude Opus 5.5: Stronger, 40% Cheaper, Tops AA Index(2026-09-23, 259 posts)
- Episode 3: Anthropic Launches Banked Usage Resets Alongside Opus 5.5(2026-09-23, 10 posts)
- Episode 4: Perplexity Rolls Out Opus 5.5 to All Users, Citing 67.6% Cost Savings(2026-09-23, 2 posts)
- Episode 5: Anthropic Teases Sonnet 5.5 and Haiku 5.5 Within Weeks(2026-09-23, 2 posts)
- Episode 6: Claude Code 2.1.280 Ships with Opus 5.5 as Default Model(2026-09-23, 3 posts)
- Episode 7: Claude Opus 5.5 Rumored to Fall Back to Weaker Models on Frontier Dev Tasks(2026-09-23, 4 posts)
- Episode 8: Claude Opus 5.5 Sparks Wave of Community Demos Within 24 Hours(2026-09-23, 14 posts)
- Episode 9: Dev recreates Minecraft in browser with Opus 5.5(2026-09-23, 2 posts)
- Episode 10: Rumor: Claude Opus 5.5 Stuns OpenAI, GPT-6.1 Timeline Squeezed(2026-09-23, 3 posts)
- Episode 11: Opus 5's FrontierCode Score Drops from 53.4% to 48% in New Card(2026-09-23, 3 posts)
- Episode 12: Developers Say Opus 5.5 Is Faster, Cheaper and Back as Daily Coding Model(2026-09-23, 4 posts)
- Episode 13: Claude Opus 5.5 Shown Generating Pixel-Art Animated Scenes(2026-09-23, 2 posts)
- Episode 14: Claude Opus 5.5 Launch Wins Rarely Unified Praise as Users Return from Codex(2026-09-24, 28 posts)
- Episode 15: Claude Opus 5.5 Tops Code Arena, Beating GPT-6 Astra by 26 Points(2026-09-24, 2 posts)
- Episode 16: Claude Opus 5.5 Delivers Fast, Proactive Coding That Developers Say Enables Flow State(2026-09-24, 2 posts)
- Opus 5 scored 53.4% on FrontierCode at launch, but its new card shows just 48% — andrew_n_carr · 2026-09-23
- Opus 5 Scored 53.4% on FrontierCode at Launch, Just 48% on the 5.5 Card — mathemagic1an · 2026-09-23
- Mystery solved: Opus 5 vs 5.5 chart gap came from thinking levels, not regression — andrew_n_carr · 2026-09-23