Claude Fable 5.1 vs Gemini 2.5 Pro Preview 0325
完整资料:Claude Fable 5.1(Anthropic)和 Gemini 2.5 Pro Preview 0325(Google),包括全部基准测试、价格和排名。
更新于 · 数据来自 12 个来源
截至 2026年11月22日,质量上 Claude Fable 5.1 领先 Gemini 2.5 Pro Preview 0325(BenchLeader Index 66.9 比 59.7)。在两者都有成绩的所有类别中,Claude Fable 5.1 均领先。
要点(截至 2026年11月22日)
- Claude Fable 5.1 的 BenchLeader Index 为 66.9,Gemini 2.5 Pro Preview 0325 为 59.7(BenchLeader,2026年11月22日)。
- Claude Fable 5.1 每百万 token 混合价格为 $20.00(标价,2026年11月22日)。
- 在两者都有结果的 5 项基准测试中,Claude Fable 5.1 领先 5 项(BenchLeader,2026年11月22日)。
按用途怎么选
推荐模型的数值列在前面。类别得分采用 BenchLeader 标尺(50 为模型平均水平,15 分为一个标准差)。领先半个标准差及以上为“高”置信度;相差不到 1 分则视为“均可”。
主要差异
- 质量上 Claude Fable 5.1 领先:66.9 比 59.7。
- Claude Fable 5.1 在编程、推理方面更强。
概览
| 规格 | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 |
|---|---|---|
| 提供商 | Anthropic | |
| 发布日期 | 2026年9月1日 | 2025年3月25日 |
| 上下文窗口(token) | 1,000,000 | 1,000,000 |
| 对比时的配置 | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 |
| 开放权重 | 否 | 否 |
| 每百万 token 价格(输入 / 输出 / 缓存输入) | $10.00 / $50.00 / $0.250 | – |
逐类别对比
类别得分采用 BenchLeader 标尺:50 为参评模型的平均水平,15 分为一个标准差。差距不到 1 分视为持平。
| 类别 | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 | 领先者 |
|---|---|---|---|
| 推理 | 73.4 | 52.1 | Claude Fable 5.1 |
| 编程 | 88.6 | 61.1 | Claude Fable 5.1 |
| 智能体与工具 | 64.1 | – | 无法比较 |
| 数学 | – | 58.9 | 无法比较 |
| 知识 | – | 56.2 | 无法比较 |
| 指令遵循 | – | 57.9 | 无法比较 |
| 多模态 | – | 70.8 | 无法比较 |
| 长上下文 | – | 72.8 | 无法比较 |
同一图表对比
各领域采用与 Index 相同的标尺,平均模型为 50 分。单击某个领域可查看构成它的基准测试。 可从列表中最多选择 5 个模型。
虚线:该模型在所经过的轴上没有结果,线条只连接相邻的点。
未绘制(这些模型没有结果):人类偏好。
各来源的分歧
在两者都有结果的 5 项基准测试中,Claude Fable 5.1 领先 5 项,Gemini 2.5 Pro Preview 0325 领先 0 项。所有同时评测两者的综合评分都认为 Claude Fable 5.1 更强。
| 综合 Index | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 | 领先者 |
|---|---|---|---|
| BenchLeader Index | 66.9 | 59.7 | Claude Fable 5.1 |
各发布方使用各自的任务、评测框架和推理强度设置,因此总体结论可能不同;BenchLeader 将它们统一到同一标尺上。† 表示同一模型在另一推理强度设置下的测量值(发布方未测试本页对比的设置)。
价格、速度与延迟
标价按每 1 个输出 token 对应 3 个输入 token 混合计算。每项任务的价格采用方法论页面中的 token 数量。每项基准测试任务的成本、token 数和耗时均为实测值,取自 Artificial Analysis 整个 Intelligence Index 测试的平均值。
| 指标 | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 |
|---|---|---|
| 每百万 token 混合价格 | $20.00 | – |
| 每百万 token 输入 / 输出 | $10.00 / $50.00 | – / – |
| 每百万 token 缓存输入 | $0.250 | – |
| 每项任务价格:聊天回复 | $0.019 | – |
| 每项任务价格:总结一份 30 页的报告 | $0.150 | – |
| 每项任务价格:代码编辑 | $0.135 | – |
| 每项任务价格:智能体编程会话 | $0.800 | – |
| 每项任务价格:结构化信息提取 | $0.030 | – |
| 智能体编程会话(输入大多命中缓存) | $0.361 | – |
| 输出速度 | 66 tok/s | – |
| 首次回答时间 | 3.8 秒 | – |
每个模型都按其排名最高的推理强度设置进行对比。Claude Fable 5.1 得分 66.9,首个回答 token 需 4 秒;Claude Fable 5.1 (max) 为 66.4(134 秒),Claude Fable 5.1 (xhigh) 为 63.9(32 秒)。Gemini 2.5 Pro Preview 0325 得分 59.7。推理强度越高,回答前思考越久:得分更高,但等待更长、计费 token 也更多。 推理强度如何影响得分和成本
全部基准测试
共 185 项基准测试,按类别分组;点击名称可查看该排行榜及领先模型。† 表示同一模型在另一推理强度设置下的测量值,仅供参考,不计入胜场。
| 指标 | 发布方 | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 66.9 | 59.7 |
| 智能体与工具得分 | BenchLeader | 64.1 | – |
| 编程得分 | BenchLeader | 88.6 | 61.1 |
| 推理得分 | BenchLeader | 73.4 | 52.1 |
| 指令遵循得分 | BenchLeader | – | 57.9 |
| 知识得分 | BenchLeader | – | 56.2 |
| 长上下文得分 | BenchLeader | – | 72.8 |
| 数学得分 | BenchLeader | – | 58.9 |
| 多模态得分 | BenchLeader | – | 70.8 |
| 混合价格(美元/百万 token) | BenchLeader | $20.00 | – |
| 输出速度 | BenchLeader | 66 tok/s | – |
| 首次回答时间 | BenchLeader | 3.8 秒 | – |
| 上下文窗口 | BenchLeader | 1,000,000 token | 1,000,000 token |
| 推理 | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | – | 83.8% |
| Humanity's Last Exam | Scale AI / CAIS | 46.5%† | 18.2% |
| SimpleBench | SimpleBench | 86.6% | 51.6% |
| Arena Hard Prompts | Arena (LMArena) | 1522† | – |
| LiveBench 2.0 Reasoning | LiveBench | 91.7%† | – |
| LiveBench 1.0 Reasoning | LiveBench | 91.7%† | – |
| GPQA Diamond (AA) | Artificial Analysis | 93.7%† | 83.6% |
| Humanity's Last Exam (AA) | Artificial Analysis | 59.1%† | 18.0% |
| HLE-Diamond (AA) | Artificial Analysis | 60.1%† | – |
| GPQA Diamond (Vals) | Vals AI | 93.4%† | 80.8% |
| GPQA Diamond (HELM) | HELM Capabilities | – | 74.9% |
| ARC-AGI-1 | ARC Prize | 97.5%† | – |
| ARC-AGI-2 | ARC Prize | 90.0%† | – |
| CritPt | Artificial Analysis | 29.7%† | – |
| CritPt v2 hard subset (AA) | Artificial Analysis | 52.5%† | – |
| MysteryMechanism | Vals AI | 47.8%† | – |
| HLE-Diamond (Scale) | Scale AI SEAL | 51.3%† | – |
| HLE Text-Only (Scale) | Scale AI SEAL | 46.8%† | 18.4% |
| Arena Expert | Arena (LMArena) | 1533† | – |
| Chess Puzzles | Epoch AI Benchmarking Hub | 48.0%† | – |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 58.0%† | – |
| LMCA | Epoch AI Benchmarking Hub | 65.5% | – |
| DTBench | Epoch AI Benchmarking Hub | 97.6% | – |
| ForecastBench | ForecastBench | 65.8%† | 64.0% |
| 编程 | |||
| SciCode | SciCode | 63.1%† | – |
显示全部 185 项基准测试
| 指标 | 发布方 | Claude Fable 5.1 | Gemini 2.5 Pro Preview 0325 |
|---|---|---|---|
| 编程 | |||
| WeirdML | WeirdML | 92.9%† | – |
| FrontierCode | Cognition | 50.9%† | – |
| GSO-Bench | GSO-Bench | 88.2% | – |
| Arena Coding | Arena (LMArena) | 1533† | – |
| Arena WebDev | Arena (LMArena) | 1742† | – |
| LiveBench 2.0 Coding | LiveBench | 89.7%† | – |
| LiveBench 1.0 Coding | LiveBench | 86.4%† | – |
| LiveCodeBench | Vals AI | 90.5%† | 79.2% |
| IOI | Vals AI | 90.8%† | – |
| Aider Polyglot | Aider polyglot leaderboard | – | 72.9% |
| Code Migration | Vals AI | 54.6%† | – |
| ProgramBench | Vals AI | 7.0%† | – |
| SWE-bench Multimodal | Vals AI | 61.3%† | – |
| Vibe Code Bench 1-100 | Vals AI | 28.0%† | – |
| Vibe Code Bench v1.1 | Vals AI | 90.3%† | – |
| SWE Atlas v1.0: Codebase QnA | Scale AI SEAL | 60.0%† | – |
| SWE Atlas v1.0: Refactoring | Scale AI SEAL | 56.7%† | – |
| SWE Atlas v1.0: Test Writing | Scale AI SEAL | 67.0%† | – |
| SWE Atlas v1.1 | Scale AI SEAL | 76.5%† | – |
| SWE Atlas v1.1: Codebase QnA | Scale AI SEAL | 79.0%† | – |
| SWE Atlas v1.1: Test Writing | Scale AI SEAL | 78.9%† | – |
| SWE Atlas v1.1: Refactoring | Scale AI SEAL | 71.4%† | – |
| SWE-Bench Pro V2 | Scale AI SEAL | 99.1%† | – |
| SWE-Bench Pro V2 HARD | Scale AI SEAL | 92.2%† | – |
| MirrorCode | Epoch AI Benchmarking Hub | 93.3%† | – |
| CadEval | Epoch AI Benchmarking Hub | – | 64.0% |
| ALE-Bench | Epoch AI Benchmarking Hub | 2143.2† | – |
| FrontierSWE | Epoch AI Benchmarking Hub | 56.3%† | – |
| WeirdML v3 | WeirdML | 26.0%† | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 52.0%† | – |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 91.4%† | – |
| Startrise Frontend | Startrise | 77.1%† | – |
| GameBench 2 Overall | DuelLab GameBench 2 | 87.0 | – |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 93.8† | – |
| Design Arena Website | Design Arena | 1298 | – |
| Design Arena Code | Design Arena | 1323 | – |
| FrontierCode 1.1 (AA) | Artificial Analysis | 49.0%† | – |
| FrontierCode 2.0 (AA) | Artificial Analysis | 31.6%† | – |
| LiveCodeBench (AA) | Artificial Analysis | – | 77.8% |
| 智能体与工具 | |||
| Terminal-Bench | Terminal-Bench | 57.3%† | – |
| Remote Labor Index | Scale AI / CAIS | 17.9% | – |
| APEX-Agents | Mercor | 68.6% | – |
| Arena Agent | Arena (LMArena) | 13.7† | – |
| LiveBench Long Agentic | LiveBench | 74.2%† | – |
| LiveBench Browser | LiveBench | 89.8%† | – |
| LiveBench Agentic Coding | LiveBench | 66.1%† | – |
| Terminal-Bench 2.1 (Vals) | Vals AI | 85.0%† | – |
| MCP Atlas | Scale AI SEAL | 87.2% | – |
| HiL-Bench | Scale AI SEAL | 61.5% | – |
| GDPval-AA v2.1, normalised score | Artificial Analysis | 62.7%† | – |
| GDPval-AA v2.1 | Artificial Analysis | 1765† | – |
| τ³-Banking (AA) | Artificial Analysis | 47.2%† | – |
| ITBench SRE (AA) | Artificial Analysis | 49.5%† | – |
| Analyst Agent (AA) | Artificial Analysis | 57.5%† | – |
| CUA-bench | Vals AI | 13.2%† | – |
| CyberBench | Vals AI | 70.4%† | – |
| Finance Agent v2 | Vals AI | 58.9%† | – |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 6.7%† | – |
| Legal Research Bench | Vals AI | 55.3%† | – |
| Robo-bench | Vals AI | 5.8%† | – |
| SkillsBench | Vals AI | 61.5%† | – |
| SREBench | Vals AI | 22.9%† | – |
| Tax Agent Bench | Vals AI | 50.8%† | – |
| Terminal-Bench 4.0 (Vals) | Vals AI | 58.1%† | – |
| Terminal-Bench Science | Vals AI | 40.0%† | – |
| Time Horizon Index: KSP | Vals AI | 63.3%† | – |
| Arena Search | Arena (LMArena) | 1227† | – |
| Arena Agent Task Completion | Arena (LMArena) | 14.2† | – |
| Arena Search Factuality | Arena (LMArena) | 1214† | – |
| BALROG | BALROG | – | 43.3% |
| PostTrainBench | Epoch AI Benchmarking Hub | 44.6%† | – |
| Vending-Bench 2 | Andon Labs | 5421.6 | – |
| SpireClimb | Epoch AI Benchmarking Hub | 2.0%† | – |
| AutomationBench | Artificial Analysis | 59.4%† | – |
| GDP.pdf (AA) | Artificial Analysis | 26.2%† | – |
| Harvey LAB-AA v1.1 | Harvey | 6.4%† | – |
| AA-Briefcase v1.1 | Artificial Analysis | 1670† | – |
| GDPx | GDPx by SemiAnalysis | 8.5%† | – |
| Design Arena Agentic | Design Arena | 1329 | – |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 43.3%† | – |
| CWE-Bench (AA) | Artificial Analysis | 50.8%† | – |
| DeepSec-Bench (AA) | Artificial Analysis | 23.5%† | – |
| CyberGym end-to-end (AA) | Artificial Analysis | 0.8%† | – |
| 数学 | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 90.2%† | – |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 87.8%† | – |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 100.0%† | – |
| MATH Level 5 | Epoch AI Benchmarking Hub | – | 95.6% |
| ProofBench | Vals AI | 100.0%† | – |
| LiveBench Mathematics | LiveBench | 97.0%† | – |
| AIME (Vals) | Vals AI | – | 85.8% |
| Omni-MATH (HELM) | HELM Capabilities | – | 41.6% |
| MATH 500 | Vals AI | – | 95.2% |
| Arena Maths | Arena (LMArena) | 1530† | – |
| FrontierMath Erdős | Epoch AI Benchmarking Hub | 0.0%† | – |
| 知识 | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 70.8%† | – |
| LiveBench 2.0 Data Analysis | LiveBench | 80.3%† | – |
| LiveBench Research | LiveBench | 88.5%† | – |
| LiveBench 1.0 Data Analysis | LiveBench | 80.3%† | – |
| AA-Omniscience v1.1 | Artificial Analysis | 46.3† | – |
| MMLU-Pro | Vals AI | 92.4%† | 84.1% |
| LegalBench | Vals AI | 88.5%† | 84.3% |
| CorpFin | Vals AI | – | 59.8% |
| TaxEval | Vals AI | 76.0%† | 72.9% |
| MedQA | Vals AI | – | 93.1% |
| PRBench Finance | Scale AI SEAL | 50.8%† | – |
| PRBench Legal | Scale AI SEAL | 51.6%† | – |
| MMLU-Pro (HELM) | HELM Capabilities | – | 86.3% |
| Excel Modeling Benchmark | Vals AI | 76.7%† | – |
| Excel Modeling Benchmark v1.1 | Vals AI | 76.7%† | – |
| MedCode | Vals AI | 53.5%† | – |
| MedScribe | Vals AI | 91.3%† | – |
| Public Benefits Bench v1.1 | Vals AI | 74.9%† | – |
| Financial Coaching Bench | Vals AI | 75.7%† | – |
| Immigration Bench | Vals AI | 75.4%† | – |
| Arena Text Factuality | Arena (LMArena) | 1499† | – |
| EBR-bench | Epoch AI Benchmarking Hub | 57.1%† | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 49.4%† | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 68.4%† | – |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 30.0%† | – |
| HealthBench Professional (AA) | OpenAI | 57.8%† | – |
| 指令遵循 | |||
| LiveBench Language | LiveBench | 89.5%† | – |
| LiveBench Instruction Following | LiveBench | 73.0%† | – |
| IFEval (HELM) | HELM Capabilities | – | 84.0% |
| WildBench (HELM) | HELM Capabilities | – | 85.7% |
| Arena Instruction Following | Arena (LMArena) | 1503† | – |
| 人类偏好 | |||
| Arena Text | Arena (LMArena) | 1503† | – |
| EQ-Bench Creative Writing v3 | EQ-Bench | 2155 | 1396 |
| EQ-Bench Longform Writing | EQ-Bench | 85.3 | 58.8 |
| Arena Creative Writing | Arena (LMArena) | 1492† | – |
| Arena Multi-turn | Arena (LMArena) | 1483† | – |
| Arena Longer Queries | Arena (LMArena) | 1511† | – |
| Lech Mazur Creative Writing | Lech Mazur | – | 80.5% |
| 多模态 | |||
| Arena Vision | Arena (LMArena) | 1322† | – |
| LiveBench Design | LiveBench | 81.6%† | – |
| VISTA | Scale AI SEAL | – | 54.6% |
| MMMU-Pro (Vals) | Vals AI | 90.6%† | 81.3% |
| MortgageTax | Vals AI | 70.8%† | 67.2% |
| Humanity's Sixth Sense | Scale AI SEAL | 40.8%† | – |
| GeoBench | GeoBench | 84.0%† | 84.0% |
| VPCT | Epoch AI Benchmarking Hub | – | 48.0% |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 41.9% | – |
| Furniture Assembly | Epoch AI Benchmarking Hub | 70.0%† | – |
| Code2Video-AA | Artificial Analysis | 1325† | – |
| AA-3D | Artificial Analysis | 1139† | – |
| 长上下文 | |||
| Fiction.LiveBench 120k | Fiction.live | – | 90.6% |
| AA-LCR v1.1 | Artificial Analysis | 85.3%† | – |
| Arena Document | Arena (LMArena) | 1498† | – |
| MLCR | Artificial Analysis | 15.6%† | – |
| 综合 | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 167.1 | 144.0 |
| LiveBench 2.0 | LiveBench | 84.0%† | – |
| LiveBench 1.0 | LiveBench | 83.4%† | – |
| AA Intelligence Index v5.0.1 | Artificial Analysis | 49.2† | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 53.4† | 15.0 |
| AA Progress Index v1 | Artificial Analysis | 340.8† | 245.9 |
| HELM Capabilities mean | HELM Capabilities | – | 74.5% |
| Vals RSI Index | Vals AI | 36.1%† | – |
| AA Cyber Index | Artificial Analysis | 25.0† | – |
| 安全与诚实 | |||
| FORTRESS | Scale AI SEAL | 13.7% | 54.9% |
| MASK | Scale AI SEAL | – | 55.9% |
| DistressBench | Scale AI SEAL | 71.5% | – |
数据截至 2026年11月22日。对比的是每个模型的最佳配置;每项分数的来源链接都在模型页面上。
DuelLab GameBench 2, release gb2-20261119-1d7fd529faf6 (https://benchmarks.duellab.org/releases/gb2-20261119-1d7fd529faf6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
变化
- Claude Fable 5.1:BenchLeader Index 66.2 → 66.9。
- Claude Fable 5.1:1 项分数修订。
- Claude Fable 5.1:1 项分数修订。
- Claude Fable 5.1:1 项新结果(SimpleBench)。
- Claude Fable 5.1:6 项分数修订。
- Claude Fable 5.1:1 项分数修订。
- Claude Fable 5.1:1 项分数修订。
- Claude Fable 5.1:3 项分数修订。
- Claude Fable 5.1:3 项新结果(Excel Modeling Benchmark v1.1, Harvey's Legal Agent Benchmark v1, SWE-bench Multimodal)。
相关对比
- Claude Fable 5.1 vs Claude Fable 5.2
- Claude Fable 5.1 vs Gemini 4 Carbon
- Claude Fable 5.1 vs GPT-6.2 Astra
- Claude Fable 5.1 vs Gemini 4 Neon
- Claude Fable 5 vs Gemini 2.5 Pro Preview 0325
- Claude Fable 5.1 vs GPT-6 Astra
- Claude Fable 5.1 vs Claude Opus 5.5
- Claude Fable 5.1 vs GPT-6.1 Sol
排行榜: 综合最佳大模型 · 最适合智能体的大模型 · 便宜好用的大模型
Claude Fable 5.1 vs Gemini 2.5 Pro Preview 0325:常见问题
- Claude Fable 5.1 比 Gemini 2.5 Pro Preview 0325 更好吗?
- 质量上 Claude Fable 5.1 领先:66.9 比 59.7。BenchLeader Index 综合了所有第三方质量基准测试;截至 2026年11月22日,Claude Fable 5.1 总体领先,但请结合你的用途查看各类别得分。
- 编程用 Claude Fable 5.1 还是 Gemini 2.5 Pro Preview 0325 更好?
- Claude Fable 5.1 在编程方面得分更高(类别得分 89 比 61,50 为平均水平)。
- 处理长文档哪个更好?
- 两者都支持 1,000,000 token 的上下文。
- 第三方基准测试对 Claude Fable 5.1 和 Gemini 2.5 Pro Preview 0325 的看法一致吗?
- 在两者都有结果的 5 项基准测试中,Claude Fable 5.1 领先 5 项,Gemini 2.5 Pro Preview 0325 领先 0 项。所有同时评测两者的综合评分都认为 Claude Fable 5.1 更强。
- Claude Fable 5.1 是什么时候发布的?
- Claude Fable 5.1 于 2026年9月1日 发布。