BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases8

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 5.5maxGPT-6 Astra
Drag a model to change the order.
Verdict
  • Claude Opus 5.5 (max) leads on quality: 69.2 vs 68.2.
  • Claude Opus 5.5 (max) is strongest in knowledge, long context, multimodal.
  • GPT-6 Astra is strongest in agents & tools, coding, maths, reasoning, composite.
  • Claude Opus 5.5 (max) is 2.5× cheaper than GPT-6 Astra ($8.00 vs $20.00 per 1M blended).
  • Claude Opus 5.5 (max) streams fastest at 98 tokens per second, 2.4× GPT-6 Astra.
MetricClaude Opus 5.5
max Anthropic
GPT-6 Astra
OpenAI
Quality
BenchLeader Index50 = average model69.2 — best of the models compared68.2
Reasoning72.985.6 — best of the models compared
Coding80.483.2 — best of the models compared
Agents & tools63.169.6 — best of the models compared
Maths74.077.5 — best of the models compared
Knowledge74.1 — best of the models compared55.7
Multimodal70.6–
Long context67.3–
Composite–73.7
Speed and latency
Output speed98 tok/s — best of the models compared40 tok/s
Time to first token3.53 s3.00 s — best of the models compared
Time to first answerafter any reasoning648.2 s–
Context window1M1.1M — 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$1.00
Index per dollar8.7 — best of the models compared3.4
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: time10.9 min10.5 s — 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$0.069
Summarize a 30-page report: time10.9 min18.0 s — 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$0.095
Code edit: time11.1 min40.5 s — 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$0.395
Agentic coding session: time11.5 min1.7 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$0.017
Structured extraction: time10.8 min8.0 s — best of the models compared

Chart

Quality againstYour models are ringed and labelled; the rest is the frontier.
55606570$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)GPT-6 AstraGPT-6 AstraGPT-6.1 Sol (max)Claude Fable 5.1 (max)Claude Fable 5Gemini 4 Argon (high)Claude Opus 5 (max)GPT-5.6 Sol (max)Gemini 4 Carbon (high)GPT-6.1 Terra (max)Gemini 3.7 Flash (high)Kimi K3.1 ProDeepSeek V4.1 Flash 1017 (max)GPT-5.6 Luna (xhigh)DeepSeek V4 Flash (high)GLM 5.3 FlashClaude Haiku 5.5 (xhigh)Step 3.5 FlashQwen3.5 Flash
  • Anthropic5
  • OpenAI5
  • Google3
  • DeepSeek2
  • Alibaba1
  • Moonshot AI1
  • StepFun1
  • 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 GPT-6 Astra
Reasoning
GPQA Diamond90.6%–
Humanity's Last Exam–54.8%
LiveBench 2.0 Reasoning (ref.)–92.7%
LiveBench 1.0 Reasoning (ref.)92.2%–
Humanity's Last Exam (AA) (ref.)61.4%–
HLE-Diamond (AA)67.9%–
ARC-AGI-197.5%–
ARC-AGI-291.7%–
CritPt (ref.)31.7%–
CritPt v2 hard subset (AA)61.0%–
Chess Puzzles (ref.)67.0%–
Mystery Game Puzzles (ref.)71.0%–
LMCA (ref.)68.2% — best of the models compared64.1%
DTBench (ref.)98.9% — best of the models compared97.1%
Coding
SciCode66.9%–
GSO-Bench–79.4%
Arena WebDev1809–
LiveBench 2.0 Coding (ref.)–93.7%
LiveBench 1.0 Coding (ref.)89.3%–
SciCode (AA) (ref.)66.9%–
MirrorCode (ref.)77.4%–
CursorBench (ref.)57.8%–
FrontierSWE (ref.)62.3%–
Terminal-Bench 4.0 (AA) (ref.)59.6%–
Startrise Frontend (ref.)–79.9%
FrontierCode 1.1 (AA) (ref.)52.2%–
FrontierCode 2.0 (AA)46.4%–
Agents & tools
Terminal-Bench64.8%–
Remote Labor Index–20.8%
APEX-Agents73.5% — best of the models compared64.7%
Arena Agent20.8–
LiveBench Long Agentic (ref.)–77.5%
LiveBench Browser (ref.)–91.3%
LiveBench Agentic Coding (ref.)71.7%–
HiL-Bench–65.3%
GDPval-AA v2.1, normalised score (ref.)68.3%–
GDPval-AA v2.1 (ref.)1866–
ITBench SRE (AA) (ref.)38.2%–
Analyst Agent (AA) (ref.)56.3%–
CUA-bench (ref.)14.0%–
Robo-bench (ref.)14.6%–
Vending-Bench 2 (ref.)–15514.7
AutomationBench (ref.)69.5%–
GDP.pdf (AA) (ref.)26.2%–
Harvey LAB-AA v1.1 (ref.)4.2%–
AA-Briefcase v1.1 (ref.)1800–
Terminal-Bench-Science 0.1 (AA) (ref.)59.0%–
CWE-Bench (AA)58.3%–
DeepSec-Bench (AA)26.9%–
CyberGym end-to-end (AA)0.8%–
FACTS Search (ref.)–89.9%
Maths
FrontierMath Tiers 1–391.2%–
FrontierMath Tier 495.0%–
OTIS Mock AIME100.0%–
ProofBench100.0% — best of the models compared99.0%
LiveBench Mathematics (ref.)97.1%–
Knowledge
SimpleQA Verified72.2%–
LiveBench 2.0 Data Analysis (ref.)–83.0%
LiveBench Research (ref.)–85.2%
LiveBench 1.0 Data Analysis (ref.)80.3%–
AA-Omniscience v1.148.1–
PRBench Finance–47.5%
PRBench Legal–48.4%
EBR-bench (ref.)71.4%–
GDP.pdf (ref.)30.6%–
AA-Omniscience v1.1: accuracy (ref.)65.5%–
AA-Omniscience v1.1: non-hallucination (ref.)49.7%–
HealthBench Professional (AA) (ref.)59.9%–
Instruction following
LiveBench Language (ref.)86.3%–
LiveBench Instruction Following (ref.)65.7%–
Multimodal
LiveBench Design (ref.)–82.0%
MMMU-Pro87.7%–
Blueprint-Bench 2 (ref.)–49.7%
Code2Video-AA (ref.)1536–
AA-3D (ref.)1196–
Long context
AA-LCR v1.184.7%–
MLCR (ref.)27.8%–
Composite
Epoch Capabilities Index (ref.)–168.8
LiveBench 2.0–85.6%
LiveBench 1.0 (ref.)83.2%–
AA Intelligence Index v5.0 (ref.)58.4–
AA Intelligence Index v4.3.2 (ref.)57.6–
AA Progress Index v1 (ref.)350–
Vals RSI Index (ref.)37.3%–
Vals Router Index (ref.)–65.0%
AA Cyber Index (ref.)28.7–
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.