Claude workhorse for coding agents, careful analysis, and production cost control
发布日期2026年2月17日
参数规模—
上下文长度1.0M
许可证Proprietary
知识截止2025年8月31日
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| OSWorld-Verified | general | 78.5 | 来源 |
| SWE-Atlas Codebase QnA (Claude Code) | code | 70.3 | 来源 |
| Terminal-Bench (Terminus-2) | reasoning code | 67.0 | 来源 |
| Terminal-Bench (Claude Code) | reasoning code | 63.1 | 来源 |
| Artificial Analysis Coding Agent Index | code | 49.4 | 来源 |
| Humanity's Last Exam | general | 46.8 | 来源 |
| Humanity's Last Exam | general | 34.6 | 来源 |
| SWE-Atlas Refactoring | code | 32.2 | 来源 |
| SWE-Atlas Test Writing | code | 31.8 | 来源 |
| SWE-Atlas Codebase QnA (Claude Code) | code | 31.2 | 来源 |
| SWE-Bench Pro | code | 14.9 | 来源 |
Pricing
API 价格对比
| 服务商 | 输入价 | 输出价 | 上下文 | 吞吐(tok/s) | 延迟(s) | 函数调用 | 代码执行 | 联网搜索 |
|---|---|---|---|---|---|---|---|---|
| 302.AI | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Abacus | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| AIHubMix | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Anthropic | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Auriko | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Azure | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| DaoXE | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| FreeModel | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| FrogBot | $3.00 | $15.00 | 200K | — | — | ✓ | ✗ | ✗ |
| LLM Gateway | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Neon | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| OpenCode Zen | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Pioneer | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| routing.run | $3.00 | $15.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Venice AI | $3.60 | $18.00 | 1.0M | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。