Compare/GPT-5.4 (low) vs Claude 3.5 Sonnet (June '24)

GPT-5.4 (low)vsClaude 3.5 Sonnet (June '24)

Side-by-side comparison of pricing, 12 benchmarks, and generation speed.

OpenAI

GPT-5.4 (low)

Input
$2.5/M
Output
$15/M
Speed
TTFT
Anthropic

Claude 3.5 Sonnet (June '24)

Input
$3/M
Output
$15/M
Speed
TTFT

Winner by Category

Cheaper
GPT-5.4 (low)
Faster (tok/s)
Tie
Lower Latency
Tie
Benchmarks (6-3)
GPT-5.4 (low)

Pricing Comparison

MetricGPT-5.4 (low)Claude 3.5 Sonnet (June '24)
Input ($/M tokens)$2.5$3
Output ($/M tokens)$15$15
Cost for 1M input + 100K output tokens:
GPT-5.4 (low)$4.00
Claude 3.5 Sonnet (June '24)$4.50

Speed Comparison

Speed data not available for these models.

Editorial Analysis

Verdict. GPT-5.4 (low) wins the overall benchmark matchup 6–3 across 9 overlapping categories, but raw benchmark score is only one input to the decision.

Pricing. Both models sit in the premium bracket for output-token pricing. At 1.0× the per-million-token cost, Claude 3.5 Sonnet (June '24) is meaningfully cheaper if your traffic is output-heavy (long completions, document generation, agent loops). Claude 3.5 Sonnet (June '24) makes more sense when output volume is low and absolute reasoning quality justifies the premium.

Strengths. GPT-5.4 (low) is strongest on GPQA Diamond (87%), IFBench (66%), SciCode (50%). Claude 3.5 Sonnet (June '24) leads on MMLU-Pro (75%), MATH-500 (70%), GPQA Diamond (56%).

Speed. Throughput is comparable — 0 tok/s vs 0 tok/s — so generation speed shouldn't drive your choice here. Look at the per-benchmark wins instead.

Provider. OpenAI and Anthropic sell to overlapping but distinct developer audiences: OpenAI tends to ship frontier reasoning models with premium positioning, while Anthropic often prices more aggressively. Your existing vendor relationships, billing, and SLA preferences may matter as much as the raw numbers above.

Workload cost. Workload scenarios (per million requests at 30M input + 15M output tokens): GPT-5.4 (low) costs $300.00 ($3600/year); Claude 3.5 Sonnet (June '24) costs $315.00 ($3780/year). At a smaller 5M-input/2M-output scale (single-developer tool or prototype): GPT-5.4 (low) ≈ $42.50/run, Claude 3.5 Sonnet (June '24) ≈ $45.00/run. At agent/realtime scale (200M input / 100M output per million requests): GPT-5.4 (low) ≈ $2000/run, Claude 3.5 Sonnet (June '24) ≈ $2100/run. GPT-5.4 (low) becomes more attractive at higher volume — the absolute per-token pricing difference compounds when you ship at scale.

Recommendation. Both models have legitimate use cases — the right answer depends on whether you are optimizing for benchmark ceiling, latency, or unit cost. Start with the cheaper / faster model, evaluate against your specific task, and only switch if the upgrade shows a meaningful lift.

Head-to-head deltas

  • Aggregate benchmark score (sum across 12 categories, capped at 100): GPT-5.4 (low) = 42, Claude 3.5 Sonnet (June '24) = 37. Within 15% — effectively equivalent if both meet the threshold your product requires.

Benchmark Comparison

Data from Artificial Analysis API — 12 benchmarks

Intelligence Index
39.18.3
Coding Index
26.0
Math Index
GPQA Diamond
87.1%56.0%
MMLU-Pro
75.1%
LiveCodeBench
AIME 2025
MATH-500
69.5%
Humanity's Last Exam
28.9%3.7%
SciCode
50.3%31.6%
IFBench
65.9%
TerminalBench
43.2%
GPT-5.4 (low)6 wins
3 winsClaude 3.5 Sonnet (June '24)

Frequently Asked Questions

Which is cheaper, GPT-5.4 (low) or Claude 3.5 Sonnet (June '24)?

GPT-5.4 (low) is cheaper overall. Its blended price (3:1 input/output ratio) is $5.63/M tokens vs $6.00/M for Claude 3.5 Sonnet (June '24).

Which model performs better on benchmarks?

GPT-5.4 (low) wins 6 out of 12 benchmarks compared to 3 for Claude 3.5 Sonnet (June '24). See the detailed benchmark chart above for per-category results.

Which is faster for real-time applications?

Both models have comparable generation speeds.

When should I use GPT-5.4 (low) vs Claude 3.5 Sonnet (June '24)?

Choose based on your priorities: GPT-5.4 (low) for lower cost, GPT-5.4 (low) for stronger benchmark performance, and both have comparable speed. For latency-sensitive apps, check the TTFT comparison above.