Anthropic's Unreleased Opus 5 Excels at Rust Optimization but Makes Stupid Mistakes
mertdumenci · x · 2026-08-14
The author tested Anthropic's unreleased Opus 5 model and found a stark contrast in its capabilities.
- Highlights: In specific niches like optimizing hot paths in a Rust codebase, it performs exceptionally well, even outperforming specialized models like Fable. The author calls it the smartest model they've used.
- Pain points: It still makes stupid mistakes frequently during daily use and communicates almost exclusively in jargon.
The author notes that this bizarre coexistence of cutting-edge intelligence and basic flaws feels like witnessing the "first alien artifact of scaled-up RL."
More from Models
- Elon Musk Teases Grok 4.6 Tied to Build Harness, Web Config Leaks Hidden Models — elonmusk · 2026-08-14
- Frontier Model Guide: Opus for Design, DeepSeek for Agents — sebo_gm · 2026-08-14
- Mistral Pivoting to Compute? Now Hosting Competitor Zhipu's GLM-5.2 — tengo_harambe · 2026-08-14
- Underrated Motif 3: Matches DeepSeek V4 Flash on Benchmarks — jinnyjuice · 2026-08-14
- Google exec confirms Gemini 4 is 'most ambitious pre-training run yet' — OfficialLoganK · 2026-08-14
- Anthropic's Suspicious Silence Hints at Upcoming 'Claudette' Drop — beffjezos · 2026-08-14