// head_to_head
Moondream 3 Preview vs Qwen3.8 27B
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.
Qwen3.8 27B is the cheaper of the two; neither can be ranked on quality here.
Moondream 3 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.
moondream
Moondream 3 Preview
- Blended / 1M
- $0.850
- Context
- 32K
- Released
- Oct 15, 2026
- Overall score
- Not evaluated
qwen
Qwen3.8 27B
- Blended / 1M
- $0.730
- Context
- 1M
- Released
- Aug 14, 2026
- Overall score
- 58.1
Specs and pricing
| Metric | Moondream 3 Preview | Qwen3.8 27B |
|---|---|---|
| LiveBench overall Mean of the seven LiveBench category scores, 0–100. Higher is better. | — | 58.1 |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | — | $1.53 |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $0.850 | $0.730win |
| Input price / 1M | $0.300 | $0.240win |
| Output price / 1M | $2.50 | $2.20win |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | — | $0.050 |
| Context window | 32K | 1Mwin |
| Max output tokens | 4K | 236Kwin |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Moondream 3 Preview on top, Qwen3.8 27B 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 | Moondream 3 Preview | Qwen3.8 27B |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $272.00/mo | $219.92/mo |
| RAG assistant 8K in / 600 out × 100K requests | $390.00/mo | $248.00/mo |
| Coding agent 40K in / 4K out × 20K requests | $440.00/mo | $261.60/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $487.50/mo | $395.50/mo |
| Bulk classification 500 in / 20 out × 5M requests | $1,000/mo | $725.00/mo |
Which should you pick?
You need to fit large documents in one call
Qwen3.8 27B
Wider context window — 1M against 32K.
Moondream 3 Preview vs Qwen3.8 27B FAQ
Which is better, Moondream 3 Preview or Qwen3.8 27B?
Qwen3.8 27B is the cheaper of the two; neither can be ranked on quality here. Moondream 3 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 Moondream 3 Preview cheaper than Qwen3.8 27B?
Qwen3.8 27B is cheaper. On a 3:1 input:output blend, Moondream 3 Preview lists at $0.850 per million tokens and Qwen3.8 27B at $0.730 — Qwen3.8 27B is 16% cheaper. Input and output are priced separately — Moondream 3 Preview charges $0.300 in and $2.50 out, Qwen3.8 27B charges $0.240 and $2.20 — so the model that looks cheaper flips depending on how output-heavy your workload is.
Does Moondream 3 Preview or Qwen3.8 27B have a bigger context window?
Qwen3.8 27B has the larger context window: 32K for Moondream 3 Preview against 1M for Qwen3.8 27B. 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 Moondream 3 Preview and Qwen3.8 27B support prompt caching?
Qwen3.8 27B publishes a cached-input rate of $0.050 per million tokens against a full input rate of $0.240. The catalogue lists no separate cached rate for Moondream 3 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.
- Scores — LiveBench release 2026-09-02, 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.