Compare / Mercury 3 vs GPT-6 Luna
Mercury 3 vs GPT-6 Luna
Which is cheaper, which is more capable, and which does better at each kind of work, from 52 results on 4 sources that measured both models.
Of the 52 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.
Results as of 19 November 2026 (catalogue release 2026-11-19-3608133e8331).
Mercury 3 or GPT-6 Luna? The short answer
Pick GPT-6 Luna: it scores higher (Spring Prompt overall 38 against 33) and costs less.
- Spring Prompt overall: Mercury 3 33, GPT-6 Luna 38 (out of 100; how the score works).
- GPT-6 Luna is cheaper to run: $0.009 against $0.012 per deck on DeckBench (26% less).
- Mercury 3 is newer: released 19 Oct 2026, against 22 Sep 2026 for GPT-6 Luna.
- On our own benchmarks, on the results both models have: GPT-6 Luna does better for decks from an analysis, user surveys and marketing planning.
Head to head
Every fact side by side
| Mercury 3 | GPT-6 Luna | |
|---|---|---|
| Price per million tokens | Not on sale | $0.10 in · $0.50 out |
| Blended (3 in : 1 out) | – | $0.20 |
| Cheapest host, blended | – | $0.20 (Amazon Bedrock) |
| Cost per deck (DeckBench) | $0.012 | $0.009 |
| Cached input, per million | – | $0.010 cached input |
| Intelligence Index | – | 38.1 |
| Spring Prompt overall | 33 | 38 |
| Output speed, tokens a second | – | 127 |
| Context window | – | 1,050,000 tokens |
| Released | 19 Oct 2026 (newer) | 22 Sep 2026 |
| Weights | Closed | Closed |
| Developer based in | the US | the US |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.
Their slides, side by side
The same brief for both: Manchester store investment review for Hearthside Coffee, an invented company. First slide of each deck from DeckBench.
From our research
- Mercury 3 first look: very fast with thinking off, too slow with it on 20 Oct 2026
- Best AI for Product Descriptions: Accuracy, Cost and Real Examples 20 Oct 2026
- Muse Ember review: GPT-6 Luna's price, better at analysis, and no Gemini Flash against a clock 10 Nov 2026
- GPT-6.1 Terra review: close to the top of our business tests at half the price of Sol 5 Nov 2026
Every shared result
Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.
| Measure | Mercury 3 | GPT-6 Luna |
|---|---|---|
| BulletBench measured by Spring Prompt · 3 results | ||
| Cost per game · Bullet 60sUS dollars, lower is better 24 · 21 of 62 measured | $0.0020no reasoning | $0.0016no reasoning |
| Ladder Elo · Bullet 60sladder Elo, higher is better 33 · 52 of 62 measured | 491no reasoning | 249no reasoning |
| Games lost on time · Bullet 60s% of games, lower is better 1 · 39 of 62 measured | 0.0%no reasoning | 50.0%no reasoning |
| DeckBench measured by Spring Prompt · 3 results | ||
| Cost per deckUS dollars, lower is better 6 · 3 of 36 measured | $0.0116 | $0.0087 |
| Deck ratingrating, higher is betterThe rating weighs each win by how strong the other deck was; the win rate counts wins against whichever decks a model happened to meet, so the two can disagree. 30 · 12 of 36 measured | 726 | 1,252 |
| Accurate decks% of tasks, higher is better 31 · 17 of 36 measured | 0.0% | 50.0% |
| ROASBench measured by Spring Prompt · 3 results | ||
| Overall scorescore out of 100, higher is better 30 · 7 of 36 measured | 26.1 | 53.4 |
| Audience scorescore out of 100, higher is better 33 · 7 of 36 measured | 43.2 | 68.3 |
| Business scorescore out of 100, higher is better 30 · 7 of 36 measured | 17.6 | 54.8 |
| SurveyBench measured by Spring Prompt · 3 results | ||
| Analysis ratingrating, higher is betterThe rating weighs each win by how strong the other analysis was; the win rate counts wins against whichever analyses a model happened to meet, so the two can disagree. 21 · 14 of 27 measured | 735 | 1,072 |
| Cost per taskUS dollars, lower is better 7 · 1 of 27 measured | $0.0216 | $0.0088 |
| Analysis scorepoints of 100, higher is better 24 · 7 of 27 measured | 42.2 | 90.8 |
Show all 52 results (40 more)
| Measure | Mercury 3 | GPT-6 Luna |
|---|---|---|
| BulletBench measured by Spring Prompt · 2 results | ||
| Invalid moves · Bullet 60s% of moves, lower is better 1 · 1 of 62 measured | 0.0%low reasoning | 0.0%no reasoning |
| Median move time · Bullet 60smilliseconds, lower is better 3 · 39 of 62 measured | 0.4 sno reasoning | 1.5 sno reasoning |
| DeckBench measured by Spring Prompt · 14 results | ||
| Caveat dropped% of tasks, lower is better 16 · 1 of 36 measured | 16.7% | 0.0% |
| Clean layout% of tasks, higher is better 10 · 10 of 36 measured | 66.7% | 66.7% |
| Design quality% of the maximum, higher is better 25 · 10 of 36 measured | 58.5% | 78.1% |
| Draft figure quoted% of tasks, lower is better 21 · 21 of 36 measured | 33.3% | 33.3% |
| Findings missing% of tasks, lower is better 25 · 1 of 36 measured | 16.7% | 0.0% |
| Head-to-head win rate% of comparisons, higher is betterThe rating weighs each win by how strong the other deck was; the win rate counts wins against whichever decks a model happened to meet, so the two can disagree. 30 · 9 of 36 measured | 35.2% | 70.9% |
| Layout defects% of tasks, lower is better 10 · 10 of 36 measured | 33.3% | 33.3% |
| Misleading metric used% of tasks, lower is better 31 · 1 of 36 measured | 16.7% | 0.0% |
| Presentable decks% of tasks, higher is better 17 · 17 of 36 measured | 0.0% | 0.0% |
| Recommendation late or wrong% of tasks, lower is better 33 · 17 of 36 measured | 83.3% | 16.7% |
| Slides needing work% of tasks, lower is better 12 · 5 of 36 measured | 33.3% | 16.7% |
| Time per deckseconds, lower is better 12 · 17 of 36 measured | 83.3 | 101 |
| Unsupported claims% of tasks, lower is better 17 · 1 of 36 measured | 16.7% | 0.0% |
| Unsupported numbers% of tasks, lower is better 28 · 1 of 36 measured | 16.7% | 0.0% |
| ROASBench measured by Spring Prompt · 6 results | ||
| Consistency scorescore out of 100, higher is better 31 · 10 of 36 measured | 15.1 | 35.8 |
| Contribution profitsimulated US dollars, higher is better 30 · 9 of 36 measured | $104,868 | $738,879 |
| Cost of a runUS dollars, lower is better 8 · 2 of 36 measured | $0.0533 | $0.0142 |
| Months over budgetmonths of 12, lower is better 1 · 1 of 36 measured | 0 | 0 |
| Planning scorescore out of 100, higher is better 35 · 20 of 36 measured | 53.8 | 54.8 |
| Return on ad spendprofit per $1 spent, higher is better 30 · 10 of 36 measured | $0.10 | $0.66 |
| SurveyBench measured by Spring Prompt · 18 results | ||
| Head-to-head win rate% of comparisons, higher is betterThe rating weighs each win by how strong the other analysis was; the win rate counts wins against whichever analyses a model happened to meet, so the two can disagree. 18 · 17 of 27 measured | 42.4% | 47.1% |
| Leading or double-barrelled questions% of tasks, lower is better 21 · 5 of 27 measured | 55.6% | 11.1% |
| Mistaken next steps% of tasks, lower is better 13 · 5 of 27 measured | 22.2% | 11.1% |
| Numbers right% of answers, higher is better 27 · 21 of 27 measured | 63.3% | 98.0% |
| Objectives unanswerable% of tasks, lower is better 1 · 1 of 27 measured | 0.0% | 0.0% |
| Output tokens per tasktokens, lower is better 15 · 4 of 17 measured | 57,343 | 14,333 |
| Precision ignored% of tasks, lower is better 1 · 1 of 27 measured | 0.0% | 0.0% |
| Research scorepoints of 100, higher is better 23 · 7 of 27 measured | 56.8 | 91.7 |
| Sound research% of tasks, higher is better 17 · 6 of 27 measured | 0.0% | 33.3% |
| Stakeholder's view presumed% of tasks, lower is better 18 · 12 of 27 measured | 22.2% | 11.1% |
| Survey plan scorepoints of 100, higher is better 18 · 9 of 27 measured | 90.8 | 94.0 |
| Survey structure% of tasks, lower is better 12 · 12 of 27 measured | 22.2% | 22.2% |
| Time per taskseconds, lower is better 16 · 12 of 27 measured | 194 | 138 |
| Traps handled% of traps, higher is better 25 · 11 of 27 measured | 60.7% | 94.2% |
| Traps missed% of tasks, lower is better 19 · 8 of 27 measured | 88.9% | 33.3% |
| Unsupported findings% of tasks, lower is better 11 · 7 of 27 measured | 44.4% | 22.2% |
| Wrong numbers% of tasks, lower is better 21 · 14 of 27 measured | 44.4% | 11.1% |
| Wrong recommendation% of tasks, lower is better 24 · 5 of 27 measured | 77.8% | 11.1% |
Compare them with other models
Quick answers
Is Mercury 3 better than GPT-6 Luna?
Pick GPT-6 Luna: it scores higher (Spring Prompt overall 38 against 33) and costs less.
Which is cheaper, Mercury 3 or GPT-6 Luna?
GPT-6 Luna is cheaper to run: $0.009 against $0.012 per deck on DeckBench (26% less).
Which is better for decks from an analysis, Mercury 3 or GPT-6 Luna?
GPT-6 Luna ranks 12 of 36 models on sale for turning a finished analysis into a deck you could present as it is, against 30 for Mercury 3.
Which is better for user surveys, Mercury 3 or GPT-6 Luna?
GPT-6 Luna ranks 14 of 27 models on sale for planning a user survey and reading its results without being misled, against 21 for Mercury 3.
Which is better for marketing planning, Mercury 3 or GPT-6 Luna?
GPT-6 Luna ranks 7 of 36 models on sale for planning a year of ad spend without overspending, against 30 for Mercury 3.
Which is better for speed under pressure, Mercury 3 or GPT-6 Luna?
Mercury 3 ranks 16 of 39 models on sale for good decisions against a real clock (fast chess), against 28 for GPT-6 Luna.
Run Mercury 3 and GPT-6 Luna on your own prompt.
Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.
