// head_to_head

Claude Sonnet 4.6 vs GPT-5.4 Mini

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.

OpenRouter + LiveBenchAll comparisonsFull leaderboard

Claude Sonnet 4.6 scores higher, GPT-5.4 Mini costs less — it depends on your workload.

Claude Sonnet 4.6 is ahead by 6.6 points overall, and GPT-5.4 Mini lists 3.6× cheaper per blended million tokens. Whether 6.6 points is worth that depends on how much a wrong answer costs you. The two cost measures disagree here, which is worth knowing: GPT-5.4 Mini has the lower sticker price, but Claude Sonnet 4.6 earns each point of capability for less — $0.1561 against $0.1871 — because per-token rates do not predict how many tokens a model actually spends on a task.

anthropic

Claude Sonnet 4.6

Blended / 1M
$6.00
Context
1M
Released
Feb 17, 2026
Overall score
73.0
reasoningtool callingimage inputfile inputprompt caching

openai

GPT-5.4 Mini

Blended / 1M
$1.69
Context
400K
Released
Mar 17, 2026
Overall score
66.4
reasoningtool callingfile inputimage inputprompt caching

Specs and pricing

MetricClaude Sonnet 4.6GPT-5.4 Mini
LiveBench overall

Mean of the seven LiveBench category scores, 0–100. Higher is better.

73.0win66.4
Cost per point

Measured benchmark spend divided by overall score — dollars per point of capability.

$0.1561win$0.1871
Blended price / 1M

3:1 input:output mix, the usual shape of production traffic.

$6.00$1.69win
Input price / 1M$3.00$0.750win
Output price / 1M$15.00$4.50win
Cached input / 1M

Price of an input token served from the prompt cache, where the provider publishes one.

$0.300$0.075win
Context window1Mwin400K
Max output tokens128K128K

Benchmarks by category

An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Claude Sonnet 4.6 on top, GPT-5.4 Mini below, both out of 100.

Agentic codingtoo close to call
42.6
41.7
Coding
79.3
71.6
Reasoning
84.8
71.3
Mathematics
87.0
78.5
Data analysis
77.9
70.8
Language
76.1
71.0
Instruction following
63.2
59.8

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.

WorkloadClaude Sonnet 4.6GPT-5.4 Mini
Support chatbot

1.2K in / 400 out × 200K requests

$1725.60/mo$491.40/mo
RAG assistant

8K in / 600 out × 100K requests

$2220.00/mo$600.00/mo
Coding agent

40K in / 4K out × 20K requests

$2088.00/mo$582.00/mo
Document extraction

20K in / 1.5K out × 50K requests

$3990.00/mo$1053.75/mo
Bulk classification

500 in / 20 out × 5M requests

$7650.00/mo$1987.50/mo
Run these two through the cost calculator

Which should you pick?

You are running this at volume

Claude Sonnet 4.6

Lowest measured cost per point of capability at $0.1561 per point — the gap compounds with every request.

Quality matters more than the bill

Claude Sonnet 4.6

Highest overall LiveBench score of the two at 73.0.

You need to fit large documents in one call

Claude Sonnet 4.6

Wider context window — 1M against 400K.

You are cost-constrained

GPT-5.4 Mini

Cheaper on blended list price at $1.69 per million tokens.

Claude Sonnet 4.6 vs GPT-5.4 Mini FAQ

Which is better, Claude Sonnet 4.6 or GPT-5.4 Mini?

Claude Sonnet 4.6 scores higher, GPT-5.4 Mini costs less — it depends on your workload. Claude Sonnet 4.6 is ahead by 6.6 points overall, and GPT-5.4 Mini lists 3.6× cheaper per blended million tokens. Whether 6.6 points is worth that depends on how much a wrong answer costs you. The two cost measures disagree here, which is worth knowing: GPT-5.4 Mini has the lower sticker price, but Claude Sonnet 4.6 earns each point of capability for less — $0.1561 against $0.1871 — because per-token rates do not predict how many tokens a model actually spends on a task.

Is Claude Sonnet 4.6 cheaper than GPT-5.4 Mini?

GPT-5.4 Mini is cheaper. On a 3:1 input:output blend, Claude Sonnet 4.6 lists at $6.00 per million tokens and GPT-5.4 Mini at $1.69 — GPT-5.4 Mini is 3.6× cheaper. Input and output are priced separately — Claude Sonnet 4.6 charges $3.00 in and $15.00 out, GPT-5.4 Mini charges $0.750 and $4.50 — so the model that looks cheaper flips depending on how output-heavy your workload is.

Claude Sonnet 4.6 vs GPT-5.4 Mini: which scores higher on benchmarks?

Claude Sonnet 4.6 scores 73.0 and GPT-5.4 Mini scores 66.4 overall on LiveBench, the mean of its seven categories. That is a 6.6-point lead for Claude Sonnet 4.6. Category scores differ from the overall figure — a model can lead on reasoning and trail on coding, which the per-category table above breaks out.

Which gives better value for money, Claude Sonnet 4.6 or GPT-5.4 Mini?

Claude Sonnet 4.6. Cost per point divides the measured dollars LiveBench spent running the benchmark by the score it earned, so it captures something token pricing misses: a reasoning model can emit many times more tokens than its per-token rate suggests. Claude Sonnet 4.6 works out at $0.1561 per point and GPT-5.4 Mini at $0.1871.

Does Claude Sonnet 4.6 or GPT-5.4 Mini have a bigger context window?

Claude Sonnet 4.6 has the larger context window: 1M for Claude Sonnet 4.6 against 400K for GPT-5.4 Mini. Note that a window you can fill is not a window you should fill — retrieval quality usually degrades well before the limit, and you pay for every token you put in it.

Do Claude Sonnet 4.6 and GPT-5.4 Mini support prompt caching?

Both publish a cached-input rate: $0.300 per million for Claude Sonnet 4.6 and $0.075 for GPT-5.4 Mini, against full input rates of $3.00 and $0.750. On a workload with a long stable prefix — a system prompt, a tool schema, a retrieved corpus — that changes the economics more than the headline price does.

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.
  • ScoresLiveBench 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.