DataLearner logo

GPT-5.5vsOpus 4.7

Across 17 shared benchmarks, Opus 4.7 leads overall: GPT-5.5 wins 7, Opus 4.7 wins 10, with 0 ties and an average score difference of -3.14.

OpenAI
GPT-5.5

OpenAI · 2026-04-23 · Reasoning model

Anthropic
Opus 4.7

Anthropic · 2026-04-16 · Reasoning model

GPT-5.57 wins(41%)(59%)10 winsOpus 4.7

Benchmark scores

Grouped by capability, sorted by largest gap within each. 17 shared benchmarks.

General Knowledge

Opus 4.7 2/3
BenchmarkGPT-5.5Opus 4.7Diff
HLE13.70351 / 563Normal (No Tools) · Text only33.30184 / 563Normal (No Tools) · Text only-19.60
CritPt1.40135 / 200Normal (No Tools)5.1094 / 200Normal (No Tools)-3.70
LiveBench79.911 / 117Deep Thinking (No Tools)76.538 / 117Deep Thinking (No Tools)+3.38

Math and Reasoning

GPT-5.5 3/3
BenchmarkGPT-5.5Opus 4.7Diff
MathArena Apex80.212 / 17Extra-High (With Tools)40.629 / 17Extra-High (With Tools)+39.59
HMMT Feb 202698.481 / 10Extra-High (With Tools)93.946 / 10Extra-High (With Tools)+4.54
AIME 20261001 / 29Extra-High (With Tools)95.8310 / 29Extra-High (With Tools)+4.17

Agent Level Benchmark

Opus 4.7 2/2
BenchmarkGPT-5.5Opus 4.7Diff
Terminal Bench Hard49.2024 / 244Normal (With Tools)54.5015 / 244Normal (With Tools)-5.30
τ²-Bench - Telecom69.30139 / 264Normal (With Tools)74128 / 264Normal (With Tools)-4.70

Claw-style Agent Evaluation

Opus 4.7 1/1
BenchmarkGPT-5.5Opus 4.7Diff
PinchBench v275.5018 / 45Reported best (effort unspecified)76.0115 / 45Reported best (effort unspecified)-0.51

Coding and Software Engineer

Opus 4.7 1/1
BenchmarkGPT-5.5Opus 4.7Diff
WeirdML v267.1522 / 52Normal (With Tools)76.4013 / 52Normal (With Tools)-9.25

Commonsense Reasoning

GPT-5.5 1/1
BenchmarkGPT-5.5Opus 4.7Diff
SimpleBench6918 / 93Normal (No Tools)61.7027 / 93Normal (No Tools)+7.30

General Evaluation

Opus 4.7 1/1
BenchmarkGPT-5.5Opus 4.7Diff
GPQA Diamond77.27264 / 462Normal (No Tools)88.50110 / 462Normal (No Tools)-11.23

Instruction Following

GPT-5.5 1/1
BenchmarkGPT-5.5Opus 4.7Diff
IF Bench46.10174 / 282Normal (No Tools)43.60189 / 282Normal (No Tools)+2.50

Long Context

Opus 4.7 1/1
BenchmarkGPT-5.5Opus 4.7Diff
AA-LCR64122 / 170Normal (No Tools)75.7082 / 170Normal (No Tools)-11.70

Multimodal Understanding

Opus 4.7 1/1
BenchmarkGPT-5.5Opus 4.7Diff
MMMU-Pro71.40119 / 227Normal (No Tools)76.4073 / 227Normal (No Tools)-5

Text Embedding

GPT-5.5 1/1
BenchmarkGPT-5.5Opus 4.7Diff
Context Arena41.9398 / 126Normal (No Tools)22.12122 / 126Normal (No Tools)+19.81

Writing and Creative Capabilities

Opus 4.7 1/1
BenchmarkGPT-5.5Opus 4.7Diff
Creative Writing1,84412 / 106Normal (No Tools)1,9079 / 106Normal (No Tools)-63.60

Specs

FieldGPT-5.5Opus 4.7
PublisherOpenAIAnthropic
Release date2026-04-232026-04-16
Model typeReasoning modelReasoning model
ArchitectureDenseDense
ParametersNot availableNot available
Context length1000K1000K
Max output128K128K

API pricing

Prices use DataLearner records when available; missing fields are not inferred.

ItemGPT-5.5Opus 4.7
Text input$5 / 1M tokens$5 / 1M tokens
Text output$30 / 1M tokens$25 / 1M tokens
Cache read$0.5 / 1M tokens$0.5 / 1M tokens
Cache write$6.25 / 1M tokens$6.25 / 1M tokens

Summary

  • GPT-5.5leads in:Math and Reasoning (3/3), Commonsense Reasoning (1/1), Instruction Following (1/1), Text Embedding (1/1)
  • Opus 4.7leads in:General Knowledge (2/3), Agent Level Benchmark (2/2), Claw-style Agent Evaluation (1/1), Coding and Software Engineer (1/1), General Evaluation (1/1), Long Context (1/1), Multimodal Understanding (1/1), Writing and Creative Capabilities (1/1)

On average across the 17 shared benchmarks, Opus 4.7 scores 3.14 higher.

Largest single-benchmark gap: Creative Writing — GPT-5.5 1,844 vs Opus 4.7 1,907 (-63.60).

Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.