// head_to_head
Mercury 2 vs GPT-5.6 Sol
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.
Mercury 2 is the cheaper of the two; neither can be ranked on quality here.
Mercury 2 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.
inception
Mercury 2
- Blended / 1M
- $0.375
- Context
- 128K
- Released
- Mar 4, 2026
- Overall score
- Not evaluated
openai
GPT-5.6 Sol
- Blended / 1M
- $4.00
- Context
- 1.1M
- Released
- Jul 9, 2026
- Overall score
- 81.1
Specs and pricing
| Metric | Mercury 2 | GPT-5.6 Sol |
|---|---|---|
| LiveBench overall Mean of the seven LiveBench category scores, 0–100. Higher is better. | — | 81.1 |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | — | $0.2870 |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $0.375win | $4.00 |
| Input price / 1M | $0.250win | $2.00 |
| Output price / 1M | $0.750win | $10.00 |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | $0.025win | $0.200 |
| Context window | 128K | 1.1Mwin |
| Max output tokens | 50K | 128Kwin |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Mercury 2 on top, GPT-5.6 Sol 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 | Mercury 2 | GPT-5.6 Sol |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $103.80/mo | $1,150/mo |
| RAG assistant 8K in / 600 out × 100K requests | $155.00/mo | $1,480/mo |
| Coding agent 40K in / 4K out × 20K requests | $134.00/mo | $1,392/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $295.00/mo | $2,660/mo |
| Bulk classification 500 in / 20 out × 5M requests | $587.50/mo | $5,100/mo |
Which should you pick?
You need to fit large documents in one call
GPT-5.6 Sol
Wider context window — 1.1M against 128K.
You are cost-constrained
Mercury 2
Cheaper on blended list price at $0.375 per million tokens.
Mercury 2 vs GPT-5.6 Sol FAQ
Which is better, Mercury 2 or GPT-5.6 Sol?
Mercury 2 is the cheaper of the two; neither can be ranked on quality here. Mercury 2 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 Mercury 2 cheaper than GPT-5.6 Sol?
Mercury 2 is cheaper. On a 3:1 input:output blend, Mercury 2 lists at $0.375 per million tokens and GPT-5.6 Sol at $4.00 — Mercury 2 is 11× cheaper. Input and output are priced separately — Mercury 2 charges $0.250 in and $0.750 out, GPT-5.6 Sol charges $2.00 and $10.00 — so the model that looks cheaper flips depending on how output-heavy your workload is.
Does Mercury 2 or GPT-5.6 Sol have a bigger context window?
GPT-5.6 Sol has the larger context window: 128K for Mercury 2 against 1.1M for GPT-5.6 Sol. 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 Mercury 2 and GPT-5.6 Sol support prompt caching?
Both publish a cached-input rate: $0.025 per million for Mercury 2 and $0.200 for GPT-5.6 Sol, against full input rates of $0.250 and $2.00. 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.
- 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.