Claude Haiku 5.5 vs GPT-5.6 Sol
完整资料:Claude Haiku 5.5(Anthropic)和 GPT-5.6 Sol(OpenAI),包括全部基准测试、价格和排名。
更新于 · 数据来自 11 个来源
截至 2026年11月16日,质量上 GPT-5.6 Sol 领先 Claude Haiku 5.5(BenchLeader Index 65.3 比 56.7)。不过,Claude Haiku 5.5 更便宜,对方的价格是它的 40 倍(每百万 token 混合价格 $0.200 比 $8.00)。在两者都有成绩的所有类别中,GPT-5.6 Sol 均领先。Claude Haiku 5.5 输出更快(约 2 倍)。
要点(截至 2026年11月16日)
- Claude Haiku 5.5 的 BenchLeader Index 为 56.7,GPT-5.6 Sol 为 65.3(BenchLeader,2026年11月16日)。
- Claude Haiku 5.5 每百万 token 混合价格为 $0.200(标价,2026年11月16日)。
- GPT-5.6 Sol 每百万 token 混合价格为 $8.00(标价,2026年11月16日)。
- Claude Haiku 5.5 比 GPT-5.6 Sol 输出更快(约 2 倍;Artificial Analysis,7 日中位数,2026年11月16日)。
- 在两者都有结果的 27 项基准测试中,GPT-5.6 Sol 领先 25 项(BenchLeader,2026年11月16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”。
| 用途 | 推荐 | 理由 | 置信度 |
|---|---|---|---|
| 编程 | GPT-5.6 Sol | 类别得分 66 比 59;SciCode 57.1% 比 51.7% | 中 |
| 数学 | GPT-5.6 Sol | 类别得分 70 比 66;OTIS Mock AIME 100.0% 比 98.9% | 中 |
| 推理 | GPT-5.6 Sol | 类别得分 65 比 57;ARC-AGI-2 92.5% 比 61.3% | 中 |
| 知识 | GPT-5.6 Sol | 类别得分 64 比 49;LiveBench 1.0 Data Analysis 79.8% 比 72.8% | 高 |
| 长文档 | GPT-5.6 Sol | 类别得分 67 比 64;AA-LCR v1.1 84.0% 比 78.3% | 中 |
| 成本最低 | Claude Haiku 5.5 | 每百万 token 混合价格 $0.200 比 $8.00;智能体编程会话 $0.008 比 $0.320 | – |
| 首次回答最快 | GPT-5.6 Sol | 首个回答 token 33.5 秒(GPT-5.6 Sol (max))比 50.5 秒(Claude Haiku 5.5 (xhigh)) | – |
| 低预算大批量 | 两者都不选 | 通常更小的模型更合适:便宜好用的大模型、最快的大模型 | – |
主要差异
- 质量上 GPT-5.6 Sol (max) 领先:65.3 比 56.7。
- GPT-5.6 Sol (max) 在编程、知识、长上下文、数学、推理方面更强。
- Claude Haiku 5.5 (xhigh) 更便宜,对方的价格是它的 40 倍(每百万 token 混合价格 $0.200 比 $8.00)。
- Claude Haiku 5.5 (xhigh) 的输出速度是对方的 2.3 倍(231 比 100 token/秒)。
概览
| 规格 | Claude Haiku 5.5 | GPT-5.6 Sol |
|---|---|---|
| 提供商 | Anthropic | OpenAI |
| 发布日期 | 2026年10月7日 | 2026年7月9日 |
| 上下文窗口(token) | 1,000,000 | 1,050,000 |
| 对比时的配置 | Claude Haiku 5.5 (xhigh) | GPT-5.6 Sol (max) |
| 开放权重 | 否 | 否 |
| 每百万 token 价格(输入 / 输出 / 缓存输入) | $0.100 / $0.500 / $0.010 | $4.00 / $20.00 / $0.400 |
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 Haiku 5.5 (xhigh) | GPT-5.6 Sol (max) | 领先者 |
|---|---|---|---|
| 推理 | 57.3 | 64.7 | GPT-5.6 Sol |
| 编程 | 58.7 | 65.8 | GPT-5.6 Sol |
| 智能体与工具 | – | 67.5 | 无法比较 |
| 数学 | 66.1 | 70.0 | GPT-5.6 Sol |
| 知识 | 49.0 | 63.8 | GPT-5.6 Sol |
| 指令遵循 | – | 71.2 | 无法比较 |
| 多模态 | – | 65.9 | 无法比较 |
| 长上下文 | 63.8 | 66.7 | GPT-5.6 Sol |
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。
未绘制(测得的模型少于两个):智能体与工具, 指令遵循, 多模态, 人类偏好。
各来源的分歧
在两者都有结果的 27 项基准测试中,Claude Haiku 5.5 领先 2 项,GPT-5.6 Sol 领先 25 项。所有同时评测两者的综合评分都认为 GPT-5.6 Sol 更强。
| 综合 Index | Claude Haiku 5.5 (xhigh) | GPT-5.6 Sol (max) | 领先者 |
|---|---|---|---|
| BenchLeader Index | 56.7 | 65.3 | GPT-5.6 Sol |
| Arena Text | 1421† | 1485 | GPT-5.6 Sol |
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 Haiku 5.5 (xhigh) | GPT-5.6 Sol (max) |
|---|---|---|
| 每百万 token 混合价格 | $0.200 | $8.00 |
| 每百万 token 输入 / 输出 | $0.100 / $0.500 | $4.00 / $20.00 |
| 每百万 token 缓存输入 | $0.010 | $0.400 |
| 每项任务价格:聊天回复 | $0.0002 | $0.0076 |
| 每项任务价格:总结一份 30 页的报告 | $0.0015 | $0.060 |
| 每项任务价格:代码编辑 | $0.0014 | $0.054 |
| 每项任务价格:智能体编程会话 | $0.0080 | $0.320 |
| 每项任务价格:结构化信息提取 | $0.0003 | $0.012 |
| 智能体编程会话(输入大多命中缓存) | $0.0040 | $0.158 |
| 每项基准测试任务成本(AA Index) | $0.248 | – |
| 每项任务输出 token(AA Index) | 136k | – |
| 其中推理部分 | 97k | – |
| 每项任务耗时(AA Index) | 10 min | – |
| 输出速度 | 231 tok/s | 100 tok/s |
| 首次回答时间 | 50.5 秒 | 33.5 秒 |
| 完整回答耗时(最近一次测量 2026年11月13日) | 110.1 秒 | 157.3 秒 |
每个模型都按其排名最高的推理强度设置进行对比。Claude Haiku 5.5 (xhigh) 得分 56.7,首个回答 token 需 50 秒;Claude Haiku 5.5 (high) 为 55.3(18 秒),Claude Haiku 5.5 (max) 为 55.3(185 秒)。GPT-5.6 Sol (max) 得分 65.3,首个回答 token 需 34 秒;GPT-5.6 Sol (xhigh) 为 63.9(17 秒),GPT-5.6 Sol (high) 为 63.4(10 秒)。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 191 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 Haiku 5.5 (xhigh) | GPT-5.6 Sol (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 56.7 | 65.3 |
| 编程得分 | BenchLeader | 58.7 | 65.8 |
| 知识得分 | BenchLeader | 49.0 | 63.8 |
| 长上下文得分 | BenchLeader | 63.8 | 66.7 |
| 数学得分 | BenchLeader | 66.1 | 70.0 |
| 推理得分 | BenchLeader | 57.3 | 64.7 |
| 智能体与工具得分 | BenchLeader | – | 67.5 |
| 指令遵循得分 | BenchLeader | – | 71.2 |
| 多模态得分 | BenchLeader | – | 65.9 |
| 混合价格(美元/百万 token) | BenchLeader | $0.200 | $8.00 |
| 输出速度 | BenchLeader | 231 tok/s | 100 tok/s |
| 首次回答时间 | BenchLeader | 50.5 秒 | 33.5 秒 |
| 上下文窗口 | BenchLeader | 1,000,000 token | 1,050,000 token |
| 推理 | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 89.7%† | 93.5% |
| Humanity's Last Exam | Scale AI / CAIS | 30.1%† | – |
| SimpleBench | SimpleBench | – | 64.8%† |
| Arena Hard Prompts | Arena (LMArena) | 1449† | 1511† |
| LiveBench 2.0 Reasoning | LiveBench | 82.4%† | – |
| LiveBench 1.0 Reasoning | LiveBench | 81.2% | 91.7% |
| GPQA Diamond (AA) | Artificial Analysis | – | 94.1% |
| Humanity's Last Exam (AA) | Artificial Analysis | 42.7% | 49.5% |
| HLE-Diamond (AA) | Artificial Analysis | 28.8% | – |
| GPQA Diamond (Vals) | Vals AI | – | 95.2% |
| EnigmaEval | Scale AI SEAL | – | 37.1%† |
| Kagi LLM Benchmark | Kagi LLM Benchmark | – | 67.0%† |
| ARC-AGI-1 | ARC Prize | 87.5% | 96.5% |
| ARC-AGI-2 | ARC Prize | 61.3% | 92.5% |
| ARC-AGI-3 | ARC Prize | 2.8%† | 7.8% |
| CritPt | Artificial Analysis | 22.6% | 32.3% |
| CritPt v2 hard subset (AA) | Artificial Analysis | 47.5% | 60.0% |
| MysteryMechanism | Vals AI | 35.1%† | 33.3% |
| HLE-Diamond (Scale) | Scale AI SEAL | – | 31.2%† |
| Arena Expert | Arena (LMArena) | 1465† | 1532† |
| Chess Puzzles | Epoch AI Benchmarking Hub | 55.0%† | 59.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 30.0%† | 58.0% |
| LMCA | Epoch AI Benchmarking Hub | 48.0% | 58.4% |
| DTBench | Epoch AI Benchmarking Hub | 87.2% | 95.5% |
| ForecastBench | ForecastBench | – | 65.7% |
显示全部 191 项基准测试
| 指标 | 发布方 | Claude Haiku 5.5 (xhigh) | GPT-5.6 Sol (max) |
|---|---|---|---|
| 推理 | |||
| BTF-3 | Epoch AI Benchmarking Hub | – | 13.7% |
| 编程 | |||
| SciCode | SciCode | 51.7% | 57.1% |
| WeirdML | WeirdML | – | 87.0% |
| FrontierCode | Cognition | 46.4%† | 47.5%† |
| GSO-Bench | GSO-Bench | – | 76.5%† |
| Arena Coding | Arena (LMArena) | 1489† | 1533† |
| Arena WebDev | Arena (LMArena) | 1593† | 1615† |
| LiveBench 2.0 Coding | LiveBench | 77.6%† | – |
| LiveBench 1.0 Coding | LiveBench | 76.4% | 83.9% |
| LiveCodeBench | Vals AI | – | 82.6% |
| IOI | Vals AI | 47.3%† | 91.2% |
| SWE-bench (Vals) | Vals AI | – | 96.2% |
| Code Migration | Vals AI | 44.6%† | 52.9% |
| Code Migration v1.1 | Vals AI | 52.4%† | – |
| ProgramBench | Vals AI | 2.5%† | 1.5% |
| SWE-bench Multimodal | Vals AI | 44.0%† | – |
| Vibe Code Bench 1-100 | Vals AI | 20.4%† | 20.0% |
| Vibe Code Bench v1.1 | Vals AI | 90.4%† | 80.5% |
| SWE Atlas v1.0: Codebase QnA | Scale AI SEAL | – | 46.0%† |
| SWE Atlas v1.0: Test Writing | Scale AI SEAL | – | 45.9%† |
| SWE Atlas v1.1 | Scale AI SEAL | 59.4%† | – |
| SWE Atlas v1.1: Codebase QnA | Scale AI SEAL | 62.1%† | – |
| SWE Atlas v1.1: Test Writing | Scale AI SEAL | 63.3%† | – |
| SWE Atlas v1.1: Refactoring | Scale AI SEAL | 52.9%† | – |
| SWE-Bench Pro V2 | Scale AI SEAL | 87.2%† | 95.5%† |
| SWE-Bench Pro V2 HARD | Scale AI SEAL | 52.9%† | 82.4%† |
| MirrorCode | Epoch AI Benchmarking Hub | 78.9%† | 20.0%† |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | – | 93.1%† |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | – | 72.7% |
| CursorBench | Cursor | 44.3% | 41.7% |
| ALE-Bench | Epoch AI Benchmarking Hub | 1587† | 2176.9 |
| FrontierSWE | Epoch AI Benchmarking Hub | 43.8%† | 32.2% |
| WeirdML v3 | WeirdML | 11.7% | 15.0%† |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 29.3% | 39.9% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 88.0% |
| Startrise Frontend | Startrise | – | 66.2%† |
| GameBench 2 Overall | DuelLab GameBench 2 | 56.0† | 67.0† |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 75.1 | 80.0† |
| Design Arena Website | Design Arena | 1253† | 1319† |
| FrontierCode 1.1 (AA) | Artificial Analysis | 45.5% | – |
| FrontierCode 2.0 (AA) | Artificial Analysis | 29.9% | – |
| 智能体与工具 | |||
| Terminal-Bench | Terminal-Bench | 40.0%† | 38.5% |
| OSWorld-Verified 2.0 | OSWorld | – | 27.3% |
| APEX-Agents | Mercor | 56.1%† | 51.4%† |
| Arena Agent | Arena (LMArena) | 0.5† | 4.6† |
| LiveBench Long Agentic | LiveBench | 47.9%† | – |
| LiveBench Browser | LiveBench | 69.5%† | – |
| LiveBench Agentic Coding | LiveBench | 51.4% | 56.2% |
| Terminal-Bench Hard | Artificial Analysis | – | 65.9% |
| τ²-Bench Telecom (AA) | Artificial Analysis | – | 85.1% |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 85.8% |
| HiL-Bench | Scale AI SEAL | – | 32.3%† |
| τ³-bench Banking | τ²-bench | – | 46.9%† |
| GDPval-AA v2.1, normalised score | Artificial Analysis | 50.7% | – |
| GDPval-AA v2.1 | Artificial Analysis | 1523 | 1619 |
| τ³-Banking (AA) | Artificial Analysis | – | 44.3% |
| ITBench SRE (AA) | Artificial Analysis | – | 56.2% |
| Analyst Agent (AA) | Artificial Analysis | 53.8%† | 47.5% |
| CUA-bench | Vals AI | – | 8.3% |
| CyberBench | Vals AI | 75.5%† | 76.3% |
| Finance Agent v2 | Vals AI | 54.1%† | 53.8% |
| Finance Agent v2.1 | Vals AI | 65.4%† | – |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | 4.2%† | – |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 1.3%† | 2.5% |
| Legal Research Bench | Vals AI | 43.8%† | 48.1% |
| SkillsBench | Vals AI | – | 54.1% |
| SREBench | Vals AI | 5.7%† | 30.5% |
| Tax Agent Bench | Vals AI | 31.9%† | 31.1% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 36.4%† | 37.9% |
| Terminal-Bench Science | Vals AI | 10.0%† | 20.0% |
| Time Horizon Index: KSP | Vals AI | – | 23.8% |
| Web Search Index | Vals AI | – | 43.6% |
| Arena Search | Arena (LMArena) | – | 1255† |
| Arena Agent Task Completion | Arena (LMArena) | 1.6† | 2.3† |
| BALROG | BALROG | – | 60.0% |
| PostTrainBench | Epoch AI Benchmarking Hub | – | 31.0% |
| GBAEval | Epoch AI Benchmarking Hub | – | 52.6%† |
| Vending-Bench 2 | Andon Labs | 3602.5† | 9619.4† |
| AutomationBench | Artificial Analysis | 36.0% | 60.1% |
| GDP.pdf (AA) | Artificial Analysis | 18.2% | 27.2% |
| Harvey LAB-AA v1.1 | Harvey | 2.8% | – |
| EnterpriseOps-Gym | Artificial Analysis | – | 42.9% |
| AA-Briefcase v1.1 | Artificial Analysis | 1523 | 1479 |
| GDPx | GDPx by SemiAnalysis | 4.9%† | – |
| Design Arena Agentic | Design Arena | – | 1238† |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 18.1% | 22.4% |
| MKA (AA) | Artificial Analysis | 41.0%† | – |
| FACTS Search | 80.3%† | 92.9%† | |
| 数学 | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 75.1%† | 89.1% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 46.3%† | 82.9% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 98.9% | 100.0% |
| ProofBench | Vals AI | 89.0%† | 83.0% |
| LiveBench Mathematics | LiveBench | 92.8% | 96.2% |
| Arena Maths | Arena (LMArena) | 1451† | 1494† |
| FrontierMath Erdős | Epoch AI Benchmarking Hub | – | 0.0% |
| 知识 | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 23.8%† | 69.7% |
| LiveBench 2.0 Data Analysis | LiveBench | 53.1%† | – |
| LiveBench Research | LiveBench | 84.3%† | – |
| LiveBench 1.0 Data Analysis | LiveBench | 72.8% | 79.8% |
| AA-Omniscience v1.1 | Artificial Analysis | 7.8 | – |
| MMLU-Pro | Vals AI | – | 89.1% |
| LegalBench | Vals AI | – | 87.0% |
| CorpFin | Vals AI | – | 64.4% |
| TaxEval | Vals AI | – | 74.8% |
| PRBench Finance | Scale AI SEAL | – | 50.5% |
| PRBench Legal | Scale AI SEAL | – | 50.5% |
| BioMysteryBench | Vals AI | 70.4%† | 71.1% |
| ClaimBench | Vals AI | 100.0%† | – |
| Excel Modeling Benchmark | Vals AI | 59.0%† | 72.3% |
| Excel Modeling Benchmark v1.1 | Vals AI | 65.5%† | – |
| MedCode | Vals AI | 51.5%† | 44.0% |
| MedScribe | Vals AI | 89.5%† | 85.2% |
| Public Benefits Bench v1.1 | Vals AI | 63.2%† | 66.5% |
| Financial Coaching Bench | Vals AI | – | 74.9% |
| Immigration Bench | Vals AI | 61.8%† | – |
| DrugDiscoveryBench | Scale AI SEAL | – | 56.5% |
| EBR-bench | Epoch AI Benchmarking Hub | 43.8%† | 44.8% |
| GDP.pdf | Epoch AI Benchmarking Hub | – | 30.7% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 33.8% | – |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 60.9% | – |
| HealthBench Professional (AA) | OpenAI | 58.4%† | – |
| FACTS Parametric | 13.1%† | 81.1%† | |
| 指令遵循 | |||
| LiveBench Language | LiveBench | 63.5% | 87.7% |
| LiveBench Instruction Following | LiveBench | 66.5% | 71.8% |
| IFBench | Artificial Analysis | – | 72.7% |
| MultiChallenge | Scale AI SEAL | 67.7%† | – |
| Arena Instruction Following | Arena (LMArena) | 1426† | 1490† |
| 人类偏好 | |||
| Arena Text | Arena (LMArena) | 1421† | 1485† |
| EQ-Bench 4 | EQ-Bench | – | 1236† |
| EQ-Bench Creative Writing v3 | EQ-Bench | – | 1974† |
| Arena Creative Writing | Arena (LMArena) | 1361† | 1471† |
| Arena Multi-turn | Arena (LMArena) | 1412† | 1491† |
| Arena Longer Queries | Arena (LMArena) | 1419† | 1498† |
| 多模态 | |||
| Arena Vision | Arena (LMArena) | – | 1282† |
| LiveBench Design | LiveBench | 79.3%† | – |
| MMMU-Pro | Artificial Analysis | – | 83.4% |
| MMMU-Pro (Vals) | Vals AI | – | 88.8% |
| MortgageTax | Vals AI | – | 67.3% |
| SAGE | Vals AI | 89.8%† | – |
| Humanity's Sixth Sense | Scale AI SEAL | – | 30.0% |
| GeoBench | GeoBench | – | 73.0%† |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 33.0%† | 33.6%† |
| Furniture Assembly | Epoch AI Benchmarking Hub | 47.5%† | 56.7% |
| VIALS-AA | Handshake AI | 49.1%† | – |
| Code2Video-AA | Artificial Analysis | 1192† | – |
| AA-3D | Artificial Analysis | 998† | – |
| FACTS Multimodal | 38.9%† | 46.4%† | |
| 长上下文 | |||
| AA-LCR v1.1 | Artificial Analysis | 78.3% | 84.0% |
| Arena Document | Arena (LMArena) | – | 1487† |
| MLCR | Artificial Analysis | 15.0% | 30.0% |
| FACTS Grounding | 57.0%† | 61.9%† | |
| 综合 | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 160.4† | 163.0† |
| LiveBench 2.0 | LiveBench | 68.9%† | – |
| LiveBench 1.0 | LiveBench | 72.1% | 81.0% |
| AA Intelligence Index v5.0.1 | Artificial Analysis | 34.6 | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 41.3 | 47.0 |
| AA Progress Index v1 | Artificial Analysis | 322.2 | 331.5 |
| Vals Index | Vals AI | 58.7† | – |
| Vals Finance Index | Vals AI | 57.5%† | – |
| Vals RSI Index | Vals AI | – | 23.9% |
| Vals Science Index | Vals AI | 38.5%† | 41.5% |
| Vals Multimodal Index | Vals AI | – | 72.6% |
| Vals Router Index | Vals AI | 58.7%† | – |
| Vals Legal Index | Vals AI | 70.3%† | – |
| 安全与诚实 | |||
| FORTRESS | Scale AI SEAL | 9.5%† | 20.1%† |
| DistressBench | Scale AI SEAL | 71.7%† | 73.6%† |
数据截至 2026年11月16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 gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
变化
- GPT-5.6 Sol (max):1 项新结果(GDPval-AA v2.1)。
- GPT-5.6 Sol (max):1 项分数修订。
- Claude Haiku 5.5 (xhigh):1 项新结果(GameBench 2 by reasoning setting)。
- GPT-5.6 Sol (max):BenchLeader Index 65.8 → 65.3。
- Claude Haiku 5.5 (xhigh):BenchLeader Index 57.3 → 56.8。
- GPT-5.6 Sol (max):1 项分数修订。
- Claude Haiku 5.5 (xhigh):BenchLeader Index 56.8 → 57.3。
- Claude Haiku 5.5 (xhigh):1 项分数修订。
- GPT-5.6 Sol (max):1 项分数修订。
- Claude Haiku 5.5 (xhigh):1 项新结果(AA-Briefcase v1.1)。
相关对比
- Claude Haiku 5.5 vs Claude Opus 5.5
- Claude Haiku 5.5 vs GPT-6 Astra
- Claude Haiku 5.5 vs Claude Fable 5.1
- GPT-5.6 Sol vs Claude Fable 5.2
- GPT-5.6 Sol vs Gemini 4 Carbon
- GPT-5.6 Sol vs Claude Opus 5.5
- GPT-5.6 Sol vs GPT-6 Astra
- Claude Sonnet 5.5 vs GPT-5.6 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 Haiku 5.5 vs GPT-5.6 Sol:常见问题
- GPT-5.6 Sol 比 Claude Haiku 5.5 更好吗?
- 质量上 GPT-5.6 Sol (max) 领先:65.3 比 56.7。BenchLeader Index 综合了所有第三方质量基准测试;截至 2026年11月16日,GPT-5.6 Sol (max) 总体领先,但请结合你的用途查看各类别得分。
- Claude Haiku 5.5 和 GPT-5.6 Sol 哪个更便宜?价格会变吗?
- 目前 Claude Haiku 5.5 更便宜:每百万 token $0.200 比 $8.00(按每 1 个输出 token 对应 3 个输入 token 混合计算)。两者均未公布价格调整。
- 编程用 Claude Haiku 5.5 还是 GPT-5.6 Sol 更好?
- GPT-5.6 Sol 在编程方面得分更高(类别得分 66 比 59,50 为平均水平)。在两者共同的最主要编程基准测试 SciCode 上,Claude Haiku 5.5 为 51.7%,GPT-5.6 Sol 为 57.1%。
- Claude Haiku 5.5 和 GPT-5.6 Sol 哪个更快?
- Claude Haiku 5.5 输出更快(约 2 倍;输出 231 比 100 token/秒,Artificial Analysis 7 日中位数)。
- 处理长文档哪个更好?
- GPT-5.6 Sol 在长上下文基准测试上得分更高(67 比 64)。两者都约为 1.1M token(Claude Haiku 5.5 为 1,000,000,GPT-5.6 Sol 为 1,050,000)。
更多问题(3)
- 哪个延迟更低?
- GPT-5.6 Sol 先开始回答:据 Artificial Analysis 测量,首个回答 token 的等待时间为 33.5 秒比 50.5 秒。这主要取决于推理强度:设置越高,回答前思考得越久。
- 第三方基准测试对 Claude Haiku 5.5 和 GPT-5.6 Sol 的看法一致吗?
- 在两者都有结果的 27 项基准测试中,Claude Haiku 5.5 领先 2 项,GPT-5.6 Sol 领先 25 项。所有同时评测两者的综合评分都认为 GPT-5.6 Sol 更强。
- Claude Haiku 5.5 是什么时候发布的?
- Claude Haiku 5.5 于 2026年10月7日 发布。