Model dossier模型档案
Claude Opus 4.8
Claude Opus 4.8 is a high-tier model from Anthropic (Intelligence Score 84.4/100). Strongest in coding (90) and reasoning (89.4); best fit: coding assistants and agentic development. It trails in preference (43.8). Premium-priced: $5.00 in / $25 out per 1M tokens. Context window 1M tokens — long-document friendly.
Claude Opus 4.8 是 Anthropic 的高水平梯队模型(智能评分 84.4/100)。 最强项是代码(90分),其次是推理(89.4分);适合代码助手与 Agent 开发。 短板是偏好(43.8分)。 定价属高端定价:每百万 token 输入 $5.00 / 输出 $25。 上下文 1M token,长文档友好。
84.4
Intelligence Score · benchmark coverage 智能评分 · benchmark 覆盖 85%
◈
Benchmark scoresBenchmark 成绩
10 entries 条| Benchmark基准 | Score成绩 | Index指数 | Dated日期 | Measured by测评方 | |
|---|---|---|---|---|---|
| Humanity's Last ExamReasoning | 46⚡max | 80 | — | 3rd-party第三方 | source ↗×6⚠±11.9 |
| GPQA DiamondReasoning | 94.3 | 99 | 2026-07-01 | 3rd-party第三方 | source ↗×4 |
| SWE-bench VerifiedCoding | 88.6 | 92 | 2026-08 | 3rd-party第三方 | source ↗×6 |
| AIME 2025Math | 98.3 | 98 | — | 3rd-party第三方 | source ↗×6 |
| LiveCodeBenchCoding | 87.82 | 94 | — | 3rd-party第三方 | source ↗×2⚠±16.62 |
| FrontierMathMath | 47.24 | 53 | — | 3rd-party第三方 | source ↗×2 |
| Terminal-BenchCoding | 74.6 | 81 | — | official官方榜 | source ↗×5⚠±10.4 |
| τ²-benchAgent | 94.4 | 95 | — | 3rd-party第三方 | source ↗×2 |
| LMArena (Chatbot Arena)Preference | 1474⚡high | 44 | — | official官方榜 | source ↗×5⚠±38 |
| WebArenaAgent | 71.2 | 77 | — | 3rd-party第三方 | source ↗ |
≡
Specs & pricing规格与价格
| Provider厂商 | Anthropic |
|---|---|
| Released发布 | 2026-05-27 |
| Context window上下文 | 1M tokens |
| Max output最大输出 | 128K tokens |
| Modality模态 | text+image+file->text |
| Reasoning tiers推理档位 | ⚡low ⚡medium ⚡high ⚡max default默认 high · low / medium / high / x-high |
| API input price输入价格 | $5.00 / 1M tokens |
| API output price输出价格 | $25 / 1M tokens |
⛓
Tools using this model使用它的工具
⚔
Head-to-head正面对比
How we score →评分方法 → · Catalog & pricing via the official OpenRouter API.模型目录与价格来自 OpenRouter 官方 API。