BenchLeader

Compare AI models side by side

As of 10 Nov 2026, the two top-ranked models are Claude Fable 5.2 (Anthropic, 72.1) and Claude Opus 5.5 (Anthropic, 68.8), 3.3 points apart on the BenchLeader Index. The cheapest model in the top 10 is GPT-6.1 Sol at $4.00 per 1M tokens blended. BenchLeader has 1073 head-to-head pages, each with a verdict, category wins, price per 1M tokens and speed.

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.

Price per 1M tokens for every model →

Claude Fable 5.2maxClaude Opus 5.5max
Drag a model to change the order.
Verdict
  • Claude Fable 5.2 (max) leads on quality: 72.1 vs 68.8.
  • Claude Fable 5.2 (max) is strongest in knowledge, maths, reasoning.
  • Claude Opus 5.5 (max) is strongest in coding, long context, agents & tools, multimodal.
  • Claude Opus 5.5 (max) is 2.0× cheaper than Claude Fable 5.2 (max) ($8.00 vs $16.00 per 1M blended).
  • Claude Fable 5.2 (max) starts answering in 3.9 min; Claude Opus 5.5 (max) takes 6.2 min.
MetricClaude Fable 5.2
max Anthropic
Claude Opus 5.5
max Anthropic
Quality
BenchLeader Index50 = average model72.1 — best of the models compared68.8
Reasoning76.0 — best of the models compared72.3
Coding80.380.5 — best of the models compared
Agents & tools–62.5
Maths74.8 — best of the models compared73.5
Knowledge86.9 — best of the models compared73.0
Multimodal–70.3
Long context66.467.1 — best of the models compared
Speed and latency
Output speed103 tok/s115 tok/s — best of the models compared
Time to first token3.94 s3.12 s — best of the models compared
Time to first answerafter any reasoning235.1 s — best of the models compared369.7 s
Context window1M — best of the models compared1M — best of the models compared
Price
Input, per 1M tokens$8.00$4.00 — best of the models compared
Output, per 1M tokens$40.00$20.00 — best of the models compared
Blended (3 in : 1 out)$16.00$8.00 — best of the models compared
Cached input, per 1Mprompt-cache read price$0.200 — best of the models compared$0.200 — best of the models compared
Index per dollar4.58.6 — best of the models compared
Cost and time per task
Chat reply: cost400 in / 300 out$0.015$0.0076 — best of the models compared
Chat reply: cost with caching75% cache hits$0.013$0.0065 — best of the models compared
Chat reply: time4.0 min — best of the models compared6.2 min
Summarize a 30-page report: cost12,000 in / 600 out$0.120$0.060 — best of the models compared
Summarize a 30-page report: cost with caching75% cache hits$0.050$0.026 — best of the models compared
Summarize a 30-page report: time4.0 min — best of the models compared6.2 min
Code edit: cost6,000 in / 1,500 out$0.108$0.054 — best of the models compared
Code edit: cost with caching75% cache hits$0.073$0.037 — best of the models compared
Code edit: time4.2 min — best of the models compared6.4 min
Agentic coding session: cost60,000 in / 4,000 out$0.640$0.320 — best of the models compared
Agentic coding session: cost with caching75% cache hits$0.289$0.149 — best of the models compared
Agentic coding session: time4.6 min — best of the models compared6.7 min
Structured extraction: cost2,000 in / 200 out$0.024$0.012 — best of the models compared
Structured extraction: cost with caching75% cache hits$0.012$0.0063 — best of the models compared
Structured extraction: time4.0 min — best of the models compared6.2 min

Chart

Quality againstYour models are ringed and labelled; the rest is the frontier.
40506070$0.100$0.200$0.500$1.00$2.00$5.00$10.00$20.00Blended price, $ per 1M tokens (log)BenchLeader IndexClaude Fable 5.2 (max)Claude Fable 5.2 (max)Claude Opus 5.5 (max)Claude Opus 5.5 (max)Gemini 4 Carbon (high)GPT-6 AstraGPT-6.1 Sol (max)Claude Fable 5.1 (max)Claude Fable 5Gemini 4 Argon (high)GPT-6.1 Terra (max)Gemini 3.7 Flash (high)DeepSeek V4 Pro (high)DeepSeek V4.1 Flash 1017 (max)GPT-5.6 Luna (xhigh)Inkling 2 SmallDeepSeek V4 Flash (high)Claude Haiku 5.5 (xhigh)Step 3.5 FlashMiMo-V2.5Muse Ember (max)Ling 3.0 FlashAgnes 2.5 Pro ThetaMistral Small 3.2Granite 4.2 8B
  • Anthropic5
  • OpenAI4
  • DeepSeek3
  • Google3
  • Ant Group1
  • IBM1
  • Meta1
  • Mistral AI1
  • Sapiens AI1
  • StepFun1
  • Thinking Machines1
  • Xiaomi1
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 Fable 5.2 max Claude Opus 5.5 max
Reasoning
GPQA Diamond93.9% — best of the models compared90.6%
LiveBench 1.0 Reasoning (ref.)–92.2%
Humanity's Last Exam (AA) (ref.)74.3% — best of the models compared61.4%
HLE-Diamond (AA)80.2% — best of the models compared67.9%
ARC-AGI-197.5% — best of the models compared97.5% — best of the models compared
ARC-AGI-292.5% — best of the models compared91.7%
ARC-AGI-394.9%–
CritPt (ref.)35.4% — best of the models compared31.7%
CritPt v2 hard subset (AA)76.5% — best of the models compared68.0%
Chess Puzzles (ref.)79.0% — best of the models compared67.0%
Mystery Game Puzzles (ref.)90.0% — best of the models compared71.0%
LMCA (ref.)–68.2%
DTBench (ref.)–98.9%
Coding
SciCode–66.9%
Arena WebDev–1818
LiveBench 1.0 Coding (ref.)–89.3%
SciCode (AA) (ref.)67.7% — best of the models compared66.9%
CursorBench (ref.)–57.8%
FrontierSWE (ref.)–62.3%
Terminal-Bench 4.0 (AA) (ref.)59.6% — best of the models compared59.6% — best of the models compared
FrontierCode 1.1 (AA) (ref.)56.2% — best of the models compared52.2%
FrontierCode 2.0 (AA)48.8% — best of the models compared46.4%
Agents & tools
Terminal-Bench–65.8%
APEX-Agents–73.5%
Arena Agent–20
LiveBench Agentic Coding (ref.)–71.7%
GDPval-AA v2.1, normalised score (ref.)–68.3%
GDPval-AA v2.1 (ref.)18311866 — best of the models compared
ITBench SRE (AA) (ref.)–38.2%
Analyst Agent (AA) (ref.)–56.3%
CUA-bench (ref.)–14.0%
Robo-bench (ref.)–14.6%
AutomationBench (ref.)76.4% — best of the models compared69.5%
GDP.pdf (AA) (ref.)24.2%26.2% — best of the models compared
Harvey LAB-AA v1.1 (ref.)7.2% — best of the models compared4.2%
AA-Briefcase v1.1 (ref.)17851801 — best of the models compared
GDPx (ref.)16.4% — best of the models compared12.0%
Terminal-Bench-Science 0.1 (AA) (ref.)73.3% — best of the models compared59.0%
CWE-Bench (AA)–58.3%
DeepSec-Bench (AA)–26.9%
CyberGym end-to-end (AA)–0.8%
Maths
FrontierMath Tiers 1–395.1% — best of the models compared91.2%
FrontierMath Tier 4100.0% — best of the models compared95.0%
OTIS Mock AIME99.7%100.0% — best of the models compared
ProofBench100.0% — best of the models compared100.0% — best of the models compared
LiveBench Mathematics (ref.)–97.1%
Knowledge
SimpleQA Verified94.0% — best of the models compared72.2%
LiveBench 1.0 Data Analysis (ref.)–80.3%
AA-Omniscience v1.166.8 — best of the models compared48.1
EBR-bench (ref.)90.5% — best of the models compared71.4%
GDP.pdf (ref.)–30.6%
AA-Omniscience v1.1: accuracy (ref.)79.1% — best of the models compared65.5%
AA-Omniscience v1.1: non-hallucination (ref.)40.7%49.7% — best of the models compared
HealthBench Professional (AA) (ref.)67.7% — best of the models compared59.9%
Instruction following
LiveBench Language (ref.)–86.3%
LiveBench Instruction Following (ref.)–65.7%
Multimodal
MMMU-Pro–87.7%
Code2Video-AA (ref.)–1536
AA-3D (ref.)1271 — best of the models compared1202
Long context
AA-LCR v1.183.3%84.7% — best of the models compared
MLCR (ref.)43.9% — best of the models compared27.8%
Composite
LiveBench 1.0 (ref.)–83.2%
AA Intelligence Index v5.0 (ref.)64.6 — best of the models compared59.1
AA Intelligence Index v4.3.2 (ref.)–57.6
AA Progress Index v1 (ref.)358 — best of the models compared350
Vals RSI Index (ref.)37.9% — best of the models compared37.3%
AA Cyber Index (ref.)–28.7
Safety & honesty
FORTRESS (ref.)–5.8%

Most-searched head-to-heads

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.