# Claude Opus 5 Claude Opus 5 is [[Anthropic]]'s Opus-tier [[Claude]] model, released 24 July 2026. API string `claude-opus-5`. See [[Claude 5]] for how the whole generation fits together. Its successor is [[Claude Opus 5.5]] (22 September 2026). It scores close to [[Claude Fable 5]] on most work at half the price, and Anthropic calls it its least prompt-injectable model yet. ## Pricing | | Input | Output | |---|---|---| | Standard | $5.00 | $25.00 | | Fast mode | 2× base | 2× base | Unchanged from Claude Opus 4.8. Fast mode gives roughly 2.5× default speed for 2× the price. ## Benchmarks Anthropic-published, so treat as directional: | Evaluation | Result | |---|---| | Frontier-Bench v0.1 | State of the art. More than double Opus 4.8's score, at lower cost per task | | CursorBench 3.2 | Within 0.5% of [[Claude Fable 5]], at half the cost | | ARC-AGI 3 | Three times the next-best model | | Zapier AutomationBench | ~1.5× the next-best pass rate | | OSWorld 2.0 | Beats Fable 5's best result at just over a third of the cost | | Organic chemistry (internal) | +10.2 points over Opus 4.8 | ## What's new - **Self-verification and longer horizons.** It checks its own work and keeps going on longer tasks - **Autonomous tool creation.** Denied direct image access during testing, it built its own computer vision pipeline. It also builds test harnesses unprompted - Mid-conversation tool changes (beta) - Automatic fallbacks to alternate models (beta) - Better visual output - Stronger on organic chemistry and protein research ## The security story [[Boris Cherny]] on prompt injection: > Opus 5 is our least prompt injectable model yet. It is a bit buried in the system card, but across PI evals and red teaming, Opus 5 is very hard to prompt inject successfully. With layered defenses (model alignment, plus prompt injection probes, plus Auto Mode in [[Claude Code]]), he reports an attack success rate of approximately zero. The number comes from the vendor. See [[Claude Code Auto Mode]]. On the safety side, Anthropic calls it their most aligned model to date with the lowest misalignment scores. Cybersecurity classifiers intervene about 85% less often than on Fable 5 while keeping vulnerability discovery ability. ## Reasoning effort levels Vals AI ran Opus 5 across all five reasoning levels on Vibe Code Bench. Performance climbs from low to medium to high, then **gets worse at the two highest levels** while costing significantly more. The regression sits within error bars: the two highest levels bring no measurable gain. On this benchmark, **high** is the highest level worth paying for. [[GPT-5.6]]'s migration guide, published the same month, also says to test one level lower than your current default. ## Reception The [[Hacker News]] thread got 1,777 points and 1,329 comments. - Commenters were impressed by the computer vision pipeline story. One pointed out that less than four years after ChatGPT, models build their own ML pipelines unprompted - One criticism was about **token burning**: when the model hits a wall, it should ask before spending your budget on an elaborate workaround - Recurring skepticism about whether it invented anything or recombined training data - Several commenters said working alongside these models **increases cognitive load** even as output goes up, because context management and code review never stop Every called it "a hard model to love." In its first week it argued with instructions, stopped before finishing work, and did not fit their existing plugin setup. According to Every, it can do impressive work but may require dismantling the systems you already built around older models. That friction comes from the context engineering changes below. ## The context engineering reset Anthropic removed **over 80% of Claude Code's system prompt** for Opus 5 and Fable 5, with no measurable loss on their coding evals. If your CLAUDE.md, skills, and prompts were tuned for older models, they may now work against the model. Run `/doctor` in [[Claude Code]]. The full set of changes is documented in [[Context Engineering]]. ## Caveats - All benchmark figures are Anthropic-published. No independent replication yet - Mid-conversation tool changes and automatic model fallbacks are both beta - The prompt injection claim comes from the vendor's own system card and red teaming - Every's experience suggests real migration friction if you have an elaborate harness ## References - Announcement: https://www.anthropic.com/news/claude-opus-5 - System card: https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf - [[Boris Cherny]] on prompt injection resistance (2026-07-24): https://x.com/bcherny/status/2080713091688583312 - Vals AI on reasoning levels (2026-07-25): https://x.com/ValsAI/status/2080817011157094511 - [[Hacker News]] discussion (1,777 points): https://news.ycombinator.com/item?id=49038433 - Every, "Vibe check: Opus 5" (paywalled): https://every.to/vibe-check/opus-5 - System prompt leak by Pliny (2026-07-25): https://x.com/elder_plinius/status/2080782804435251373 ## Related - [[Claude Opus 5.5]] - [[GPT-6 Astra]] - [[DeepSeek V4.1 Flash]] - [[Claude 5]] - [[Claude Fable 5]] - [[Claude Sonnet 5]] - [[Claude Fable 5.1]] - [[Claude Opus 4.7]] - [[Anthropic]] - [[Claude]] - [[Claude Code]] - [[Context Engineering]] - [[GPT-5.6]] - [[AI Frontier Model]] - [[Hacker News]]