The Known Good Updated 25 Jul 2026
Anthropic: Claude Opus 4.6 vs OpenAI: GPT-5.4 ·Anthropic vs OpenAI

Anthropic: Claude Opus 4.6 vs OpenAI: GPT-5.4 — head-to-head comparison

Anthropic: Claude Opus 4.6 profile OpenAI: GPT-5.4 profile All comparisons
Anthropic: Claude Opus 4.6 — Anthropic
Quality
#11 of 61
88based on 5 of 6 evaluations
Known Good Index
Speed
#219 of 316
36.7
Output tokens per second
Price
#315 of 340
$10.00
Blended per 1M tokens
In $5.00 · Out $25.00
Cache hit price
#186 of 206
$0.50
per 1M tokens (−90%)
Context
#85 of 345
1.00M
tokens (~1,500 pages)
OpenAI: GPT-5.4 — OpenAI
Quality
#10 of 61
90based on 5 of 6 evaluations
Known Good Index
Speed
#78 of 316
82.3
Output tokens per second
Price
#299 of 340
$5.62
Blended per 1M tokens
In $2.50 · Out $15.00
Cache hit price
#160 of 206
$0.25
per 1M tokens (−90%)
Context
#15 of 345
1.05M
tokens (~1,600 pages)
Summary

On the Known Good Index, Anthropic: Claude Opus 4.6 scores 88 and OpenAI: GPT-5.4 scores 90 — a 2-point lead for OpenAI: GPT-5.4. They share 5 published evaluations, with Anthropic: Claude Opus 4.6 ahead on 1 and OpenAI: GPT-5.4 ahead on 4.

Blended pricing is $10.00 per 1M tokens for Anthropic: Claude Opus 4.6 and $5.62 for OpenAI: GPT-5.4 — OpenAI: GPT-5.4 is about 1.8× cheaper.

Throughput is 37 tokens per second for Anthropic: Claude Opus 4.6 and 82 for OpenAI: GPT-5.4, with time to first token at 2.1s and 2.1s. Anthropic: Claude Opus 4.6 accepts 1.00M tokens of context against 1.05M.

Anthropic: Claude Opus 4.6 is proprietary; OpenAI: GPT-5.4 is proprietary. 5 tracked hosts serve one or both, and 1 serve both.

Generated from tracked data on 25 Jul 2026 · how these comparisons are computed →

At a glance
MetricAnthropic: Claude Opus 4.6OpenAI: GPT-5.4
Known Good Index 88 based on 5 of 6 evaluations 90 based on 5 of 6 evaluations
Blended $/1M $10.00 $5.62
Output tok/s 37 82
Context window 1.00M 1.05M
Providers 4 2
Jump to section

Quality

Known Good Index and the capability indices, side by side

IndicesComponents
Index scores

Each index is min-max normalised across tracked models and scaled 0–100 from published evaluations. We do not run these evaluations.

MetricAnthropic: Claude Opus 4.6OpenAI: GPT-5.4
Known Good Index 88 90
Coding Index 89 89
Mathematics Index 93 95
Science Index 95 98
Reasoning Index 84 88
Agentic Index 89 89
Openness Index 5 5
The Known Good

Quality Breakdown

Every evaluation both models report a published score for

Evaluations both models report

Bar = Anthropic: Claude Opus 4.6 · marker = OpenAI: GPT-5.4 · scores ingested from Epoch AI (CC-BY 4.0)

GPQA Diamond91 / 93
MATH / AIME93 / 95
SWE-bench Verified79 / 77
Humanity's Last Exam34 / 36
Terminal-Bench80 / 82

Right-hand figures read Anthropic: Claude Opus 4.6 / OpenAI: GPT-5.4. Anthropic: Claude Opus 4.6 leads on 1 of 5, OpenAI: GPT-5.4 on 4.

The Known Good

Arena Elo

Human preference rating from blind pairwise votes

Arena ratings

Elo from blind pairwise votes, ingested from Arena. Higher is better.

MetricAnthropic: Claude Opus 4.6OpenAI: GPT-5.4
Text Arena 1,497 1,453
Text Arena (style controlled) 1,498 1,466
The Known Good

Price & Cost

Published API pricing by token type

Per token typeCache discount
Published pricing

USD per 1M tokens. Blended is a 3:1 input:output mix. Live from OpenRouter, cross-checked against models.dev. Lower is better.

MetricAnthropic: Claude Opus 4.6OpenAI: GPT-5.4
Input $/1M $5.00 $2.50
Output $/1M $25.00 $15.00
Blended $/1M $10.00 $5.62
Cache read $/1M $0.500 $0.250
The Known Good
We publish blended price rather than cost-per-task. Cost-per-task requires token counts from an evaluation suite we do not run. Blended price is computed from published rates and is fully reproducible — see the pricing methodology.

Context Window

Maximum input accepted and maximum output emitted

Context and output limits

Provider-declared limits, from metadata via models.dev and OpenRouter. Higher is better.

MetricAnthropic: Claude Opus 4.6OpenAI: GPT-5.4
Context window 1.00M 1.05M
Max output tokens 128k 128k
The Known Good

Speed & Latency

Median throughput and time to first token across hosting providers

Output speedTime to first token
Throughput and latency

Medians across all tracked providers, derived from OpenRouter endpoint telemetry.

MetricAnthropic: Claude Opus 4.6OpenAI: GPT-5.4
Output tokens/s 37 82
Time to first token 2.1s 2.1s
The Known Good

Providers

Which hosts serve each model, and at what price and speed

5 hosts · 1 serve both
Full provider benchmarks →
Provider Anthropic: Claude Opus 4.6 blendedAnthropic: Claude Opus 4.6 tok/s OpenAI: GPT-5.4 blendedOpenAI: GPT-5.4 tok/s
Google · first party $10.00 40 — —
Anthropic · first party $10.00 37 — —
Amazon Bedrock · first party $10.00 33 — —
Azure · first party $10.00 30 $5.62 79
OpenAI · first party — — $2.81 68

Per-host pricing and telemetry from OpenRouter. A dash means that host does not serve that model.

The Known Good
Sources for this page
Pricing & providers — OpenRouter Metadata — models.dev Benchmarks — Epoch AI (CC-BY 4.0)Elo — Arena Full attribution →