// head_to_head

Lyria 3 Pro Preview vs Muse Spark 1.3

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

Lyria 3 Pro Preview is the cheaper of the two; neither can be ranked on quality here.

Lyria 3 Pro Preview 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.

google

Lyria 3 Pro Preview

Blended / 1M
Free
Context
1.0M
Released
Mar 30, 2026
Overall score
Not evaluated
image input

meta

Muse Spark 1.3

Blended / 1M
$2.00
Context
1.0M
Released
Sep 2, 2026
Overall score
81.6
reasoningtool callingimage inputvideo inputfile inputaudio inputprompt caching

Specs and pricing

MetricLyria 3 Pro PreviewMuse Spark 1.3
LiveBench overall

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

81.6
Cost per point

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

$0.1210
Blended price / 1M

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

Freewin$2.00
Input price / 1MFreewin$1.25
Output price / 1MFreewin$4.25
Cached input / 1M

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

$0.150
Context window1.0M1.0M
Max output tokens66K944Kwin

Benchmarks by category

An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Lyria 3 Pro Preview on top, Muse Spark 1.3 below, both out of 100.

Agentic coding
64.1
Coding
81.1
Reasoning
89.7
Mathematics
95.9
Data analysis
79.6
Language
82.8
Instruction following
78.0

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.

WorkloadLyria 3 Pro PreviewMuse Spark 1.3
Support chatbot

1.2K in / 400 out × 200K requests

$0/mo$560.80/mo
RAG assistant

8K in / 600 out × 100K requests

$0/mo$815.00/mo
Coding agent

40K in / 4K out × 20K requests

$0/mo$724.00/mo
Document extraction

20K in / 1.5K out × 50K requests

$0/mo$1,514/mo
Bulk classification

500 in / 20 out × 5M requests

$0/mo$3,000/mo
Run these two through the cost calculator

Which should you pick?

You are cost-constrained

Lyria 3 Pro Preview

Cheaper on blended list price at Free per million tokens.

Lyria 3 Pro Preview vs Muse Spark 1.3 FAQ

Which is better, Lyria 3 Pro Preview or Muse Spark 1.3?

Lyria 3 Pro Preview is the cheaper of the two; neither can be ranked on quality here. Lyria 3 Pro Preview 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 Lyria 3 Pro Preview cheaper than Muse Spark 1.3?

Lyria 3 Pro Preview is cheaper. On a 3:1 input:output blend, Lyria 3 Pro Preview lists at Free per million tokens and Muse Spark 1.3 at $2.00. Input and output are priced separately — Lyria 3 Pro Preview charges Free in and Free out, Muse Spark 1.3 charges $1.25 and $4.25 — so the model that looks cheaper flips depending on how output-heavy your workload is.

Does Lyria 3 Pro Preview or Muse Spark 1.3 have a bigger context window?

They are effectively the same — 1.0M for Lyria 3 Pro Preview and 1.0M for Muse Spark 1.3.

Do Lyria 3 Pro Preview and Muse Spark 1.3 support prompt caching?

Muse Spark 1.3 publishes a cached-input rate of $0.150 per million tokens against a full input rate of $1.25. The catalogue lists no separate cached rate for Lyria 3 Pro Preview, 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.
  • 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.