Claude Opus 5 launches with strong coding feedback and mixed benchmark signals

Latent Space · rss · 2026-07-25

Anthropic’s Claude Opus 5 launches with benchmark and user feedback that paint a mixed but strong picture.

• Epoch reports ECI 159 for Opus 5, slightly below Fable 5’s 161, while its SWE-ECI is 161, matching Fable 5 on software engineering.
• Early users describe the model as materially better in real work than public scores suggest, especially for coding and browser/tool use.
• One evaluator noted an oddity on FrontierCode: medium-effort runs outperformed higher-effort ones, hinting that more inference-time compute is not always better.
• Other reactions praise best-of-n coding results, browser automation, and practical agentic behavior, though some of that is anecdotal rather than systematic.

Original post →

More from Models

• Bug Hunt Bench ranks frontier coding models on 105 planted real-repo bugs — PawelHuryn · 2026-09-11
• GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11
• 105 hidden bugs, 2 repos: DeepSeek V4.1 Flash fixes 24 at $1.80 vs Opus 5's 27 at $51.33 — ChartsJournalX · 2026-09-11
• awesome-llm-leaderboards: an open-source directory of LLM leaderboards, pricing tables, comparison tools — Last_Establishment_1 · 2026-09-11
• Anthropic claims it works to keep eval environments unidentifiable to models — MaxKannen · 2026-09-11
• Nex N2.5 Pro released on Hugging Face with 407GB of weights — jinnyjuice · 2026-09-11

Models channel →