BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

Compare

Pick up to six models. You get a plain-language verdict, a spec sheet with the best value in each row highlighted, a chart, and every benchmark score side by side.

Claude Opus 5.5maxClaude Fable 5.1max
Drag a model to change the order.
Verdict
  • Claude Opus 5.5 (max) leads on quality: 72.4 vs 70.4.
  • Claude Opus 5.5 (max) is strongest in agents & tools, coding, composite, knowledge, maths, multimodal.
  • Claude Fable 5.1 (max) is strongest in long context, reasoning, human preference.
  • Claude Opus 5.5 (max) is 2.5× cheaper than Claude Fable 5.1 (max) ($8.00 vs $20.00 per 1M blended).
  • Claude Opus 5.5 (max) streams fastest at 98 tokens per second, 1.4× Claude Fable 5.1 (max).
  • Claude Fable 5.1 (max) starts answering in 4.0 min; Claude Opus 5.5 (max) takes 11.9 min.
MetricClaude Opus 5.5
max Anthropic
Claude Fable 5.1
max Anthropic
Quality
BenchLeader Index50 = average model72.4 — best of the models compared70.4
Reasoning75.576.6 — best of the models compared
Coding81.9 — best of the models compared74.9
Agents & tools63.2 — best of the models compared60.5
Maths74.3 — best of the models compared72.9
Knowledge78.6 — best of the models compared77.3
Human preference–70.0
Multimodal70.6 — best of the models compared67.1
Long context67.367.7 — best of the models compared
Composite93.3 — best of the models compared88.6
Speed and latency
Output speed98 tok/s — best of the models compared70 tok/s
Time to first token3.75 s3.12 s — best of the models compared
Time to first answerafter any reasoning712.3 s239.8 s — best of the models compared
Context window1M — best of the models compared1M — best of the models compared
Price
Input, per 1M tokens$4.00 — best of the models compared$10.00
Output, per 1M tokens$20.00 — best of the models compared$50.00
Blended (3 in : 1 out)$8.00 — best of the models compared$20.00
Cached input, per 1Mprompt-cache read price$0.200 — best of the models compared$0.250
Index per dollar9.1 — best of the models compared3.5
Cost and time per task
Chat reply: cost400 in / 300 out$0.0076 — best of the models compared$0.019
Chat reply: cost with caching75% cache hits$0.0065 — best of the models compared$0.016
Chat reply: time11.9 min4.1 min — best of the models compared
Summarize a 30-page report: cost12,000 in / 600 out$0.060 — best of the models compared$0.150
Summarize a 30-page report: cost with caching75% cache hits$0.026 — best of the models compared$0.062
Summarize a 30-page report: time12.0 min4.1 min — best of the models compared
Code edit: cost6,000 in / 1,500 out$0.054 — best of the models compared$0.135
Code edit: cost with caching75% cache hits$0.037 — best of the models compared$0.091
Code edit: time12.1 min4.4 min — best of the models compared
Agentic coding session: cost60,000 in / 4,000 out$0.320 — best of the models compared$0.800
Agentic coding session: cost with caching75% cache hits$0.149 — best of the models compared$0.361
Agentic coding session: time12.6 min5.0 min — best of the models compared
Structured extraction: cost2,000 in / 200 out$0.012 — best of the models compared$0.030
Structured extraction: cost with caching75% cache hits$0.0063 — best of the models compared$0.015
Structured extraction: time11.9 min4.0 min — best of the models compared

Chart

Quality againstYour models are ringed and labelled; the rest is the frontier.
506070$0.050$0.100$0.200$0.500$1.00$2.00$5.00$10.00$20.00Blended price, $ per 1M tokens (log)BenchLeader IndexClaude Opus 5.5 (max)Claude Opus 5.5 (max)Claude Fable 5.1 (max)Claude Fable 5.1 (max)GPT-6 Astra (max)GPT-6.1 Sol (max)Gemini 4 Argon (high)Claude Fable 5 (max)Claude Sonnet 5.5 (max)Claude Opus 5 (max)Muse Spark 1.3 (max)Kimi K3.1 ProGPT-6.1 Terra (max)Gemini 3.8 Flash (high)DeepSeek V4.1 Flash 1017 (max)GLM 5.3 FlashClaude Haiku 5.5 (xhigh)Step 5 PreviewMiMo-V2.6-FlashMiMo-V2.5Step 3.5 FlashLing 3.0 FlashQwen3.5-4B (thinking)Granite 4.2 3B
  • Anthropic6
  • OpenAI3
  • Google2
  • StepFun2
  • Xiaomi2
  • Alibaba1
  • Ant Group1
  • DeepSeek1
  • IBM1
  • Meta1
  • Moonshot AI1
  • Zhipu AI1
Dashed line: the frontier. Purple-ringed points are not beaten on both axes by any other model shown.

Every benchmark

Raw scores side by side, grouped by category. Best value in each row is purple. “Ref.” benchmarks are shown but not counted in the index.

Benchmark Claude Opus 5.5 max Claude Fable 5.1 max
Reasoning
GPQA Diamond90.6%–
Arena Hard Prompts–1523
LiveBench 1.0 Reasoning (ref.)92.2% — best of the models compared91.7%
GPQA Diamond (AA) (ref.)–93.7%
Humanity's Last Exam (AA) (ref.)61.4% — best of the models compared59.1%
ARC-AGI-197.5% — best of the models compared97.5% — best of the models compared
ARC-AGI-291.7% — best of the models compared90.0%
CritPt31.7% — best of the models compared29.7%
Chess Puzzles (ref.)67.0% — best of the models compared48.0%
Mystery Game Puzzles (ref.)71.0% — best of the models compared58.0%
LMCA (ref.)68.2%–
DTBench (ref.)98.9%–
Coding
SciCode66.9% — best of the models compared63.1%
WeirdML–92.9%
Arena Coding–1538
Arena WebDev1807 — best of the models compared1743
LiveBench 1.0 Coding (ref.)89.3% — best of the models compared86.4%
SciCode (AA) (ref.)66.9% — best of the models compared63.1%
MirrorCode (ref.)77.4%–
CursorBench (ref.)57.8% — best of the models compared51.8%
FrontierSWE (ref.)62.3% — best of the models compared56.3%
Terminal-Bench 4.0 (AA) (ref.)59.6% — best of the models compared52.0%
Terminal-Bench 2.1 (AA) (ref.)–91.4%
FrontierCode 1.1 (AA) (ref.)52.2% — best of the models compared49.0%
Agents & tools
Terminal-Bench64.8% — best of the models compared57.9%
APEX-Agents73.5%–
Arena Agent20.8 — best of the models compared13.9
LiveBench Agentic Coding (ref.)71.7% — best of the models compared66.1%
GDPval-AA v2.1 (ref.)68.3% — best of the models compared62.7%
τ³-Banking (AA)–47.2%
ITBench SRE (AA) (ref.)38.2%49.5% — best of the models compared
Analyst Agent (AA) (ref.)56.3%57.5% — best of the models compared
CUA-bench (ref.)14.0% — best of the models compared13.2%
Robo-bench (ref.)14.6% — best of the models compared5.8%
AutomationBench (ref.)69.5% — best of the models compared59.4%
GDP.pdf (AA) (ref.)26.2% — best of the models compared26.2% — best of the models compared
Harvey LAB-AA v1.1 (ref.)4.2%6.4% — best of the models compared
AA-Briefcase v1.1 (ref.)1800 — best of the models compared1669
Terminal-Bench-Science 0.1 (AA) (ref.)59.0% — best of the models compared43.3%
CWE-Bench (AA)58.3% — best of the models compared50.8%
DeepSec-Bench (AA)26.9% — best of the models compared23.5%
CyberGym end-to-end (AA)0.8% — best of the models compared0.8% — best of the models compared
Maths
FrontierMath Tiers 1–391.2% — best of the models compared90.2%
FrontierMath Tier 495.0% — best of the models compared87.8%
OTIS Mock AIME100.0% — best of the models compared100.0% — best of the models compared
ProofBench100.0% — best of the models compared100.0% — best of the models compared
LiveBench Mathematics (ref.)97.1% — best of the models compared97.0%
Arena Maths–1529
Knowledge
SimpleQA Verified72.2% — best of the models compared70.8%
LiveBench 1.0 Data Analysis (ref.)80.3% — best of the models compared80.3%
AA-Omniscience46.4 — best of the models compared43.5
EBR-bench (ref.)71.4% — best of the models compared57.1%
GDP.pdf (ref.)30.6% — best of the models compared27.6%
AA-Omniscience: accuracy (ref.)66.2%67.2% — best of the models compared
AA-Omniscience: non-hallucination (ref.)41.4% — best of the models compared27.4%
Instruction following
LiveBench Language (ref.)86.3%89.5% — best of the models compared
LiveBench Instruction Following (ref.)65.7%73.0% — best of the models compared
Arena Instruction Following (ref.)–1501
Human preference
Arena Text–1503
Arena Creative Writing (ref.)–1487
Arena Multi-turn (ref.)–1488
Arena Longer Queries (ref.)–1511
Multimodal
Arena Vision–1320
MMMU-Pro87.7%–
AA-3D (ref.)1196 — best of the models compared1135
Long context
AA-LCR84.7%85.3% — best of the models compared
Arena Document (ref.)–1493
MLCR (ref.)27.8% — best of the models compared15.6%
Composite
LiveBench 1.0 (ref.)83.2%83.4% — best of the models compared
AA Intelligence Index v4.3.257.6 — best of the models compared53.4
Vals RSI Index (ref.)37.3% — best of the models compared36.1%
AA Cyber Index (ref.)28.7 — best of the models compared25.0
Safety & honesty
FORTRESS (ref.)5.8%–

Demand pairs first, then every pairing of the top 15 models: each with a verdict, category wins, price and speed. Each model’s page links the rest of its comparisons.