// head_to_head
Claude Opus 4.8 vs GPT-5.4 Pro
Published LiveBench scores across all seven categories, live list pricing, context windows, and the measured cost of a point of capability — for both models, side by side.
Claude Opus 4.8 is the cheaper of the two; neither can be ranked on quality here.
GPT-5.4 Pro does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.
anthropic
Claude Opus 4.8
- Blended / 1M
- $10.00
- Context
- 1M
- Released
- May 27, 2026
- Overall score
- 76.2
openai
GPT-5.4 Pro
- Blended / 1M
- $67.50
- Context
- 1.1M
- Released
- Mar 5, 2026
- Overall score
- Not evaluated
Specs and pricing
| Metric | Claude Opus 4.8 | GPT-5.4 Pro |
|---|---|---|
| LiveBench overall Mean of the seven LiveBench category scores, 0–100. Higher is better. | 76.2 | — |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | $0.5347 | — |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $10.00win | $67.50 |
| Input price / 1M | $5.00win | $30.00 |
| Output price / 1M | $25.00win | $180.00 |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | $0.500 | — |
| Context window | 1M | 1.1M |
| Max output tokens | 128K | 128K |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Claude Opus 4.8 on top, GPT-5.4 Pro below, both out of 100.
What each one costs to run
Per-token prices are hard to feel. These are monthly list costs for both models across five workload shapes, using each provider's published cached-input rate where there is one.
| Workload | Claude Opus 4.8 | GPT-5.4 Pro |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $2,876/mo | $21,600/mo |
| RAG assistant 8K in / 600 out × 100K requests | $3,700/mo | $34,800/mo |
| Coding agent 40K in / 4K out × 20K requests | $3,480/mo | $38,400/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $6,650/mo | $43,500/mo |
| Bulk classification 500 in / 20 out × 5M requests | $12,750/mo | $93,000/mo |
Which should you pick?
You are cost-constrained
Claude Opus 4.8
Cheaper on blended list price at $10.00 per million tokens.
Claude Opus 4.8 vs GPT-5.4 Pro FAQ
Which is better, Claude Opus 4.8 or GPT-5.4 Pro?
Claude Opus 4.8 is the cheaper of the two; neither can be ranked on quality here. GPT-5.4 Pro does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.
Is Claude Opus 4.8 cheaper than GPT-5.4 Pro?
Claude Opus 4.8 is cheaper. On a 3:1 input:output blend, Claude Opus 4.8 lists at $10.00 per million tokens and GPT-5.4 Pro at $67.50 — Claude Opus 4.8 is 6.8× cheaper. Input and output are priced separately — Claude Opus 4.8 charges $5.00 in and $25.00 out, GPT-5.4 Pro charges $30.00 and $180.00 — so the model that looks cheaper flips depending on how output-heavy your workload is.
Does Claude Opus 4.8 or GPT-5.4 Pro have a bigger context window?
They are effectively the same — 1M for Claude Opus 4.8 and 1.1M for GPT-5.4 Pro.
Do Claude Opus 4.8 and GPT-5.4 Pro support prompt caching?
Claude Opus 4.8 publishes a cached-input rate of $0.500 per million tokens against a full input rate of $5.00. The catalogue lists no separate cached rate for GPT-5.4 Pro, which means the provider does not price it separately here — not that caching is unavailable.
Related comparisons
How these numbers are produced
- Price — provider list price from OpenRouter, refreshed every 15 minutes. “Blended” is a 3:1 input:output mix.
- Scores — LiveBench release 2026-06-25, using their own category map. Each model shows its strongest published run. A blank means “not evaluated”, never “bad”.
- Cost per point — the measured dollars LiveBench spent on the run, divided by the score it earned.
- “Win” — awarded only past a threshold: one full point on a benchmark score, 10% on a price, 25% on a context window. Anything tighter reports as a tie, because effort settings alone move a LiveBench score by more than that.
Published benchmarks rank models on someone else's tasks. Before committing, see LLM & agent evaluation for building an eval on your own.