Gemini 3.5 Flash Lite vs GPT-6 Luna
Same documents, same prompt: where each model gets it right, what it really costs, and how fast it responds.
Verdict
GPT-6 Luna is more accurate and cheaper in 30 of 30 combinations.
More accurate
GPT-6 Luna
wins 20 of 30
Cheaper
GPT-6 Luna
cheaper in 30 of 30
Faster
Gemini 3.5 Flash Lite
0.9 sec vs 3.0 sec
Fewer fixes
GPT-6 Luna
33 vs 67 documents per 1,000
Across 30 task and document-language combinations (speed and fixes: median).
At a glance
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference |
| Invoice extraction | 99.21% | 99.70% | ▲ 0.49 | $0.87 | $0.51 | −41 % |
| Email triage | 96.30% | 97.48% | ▲ 1.18 | $0.42 | $0.13 | −68 % |
| Personal data detection | 99.52% | 99.62% | ▲ 0.1 | $0.41 | $0.11 | −74 % |
| Contract clauses | 99.95% | 99.79% | ▼ 0.16 | $0.69 | $0.20 | −71 % |
| Document summaries | 99.22% | 100.00% | ▲ 0.78 | $0.46 | $0.16 | −66 % |
| Questions about a document | 100.00% | 99.11% | ▼ 0.89 | $0.46 | $0.15 | −67 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference |
| Invoice extraction | 99.49% | 99.28% | ▼ 0.21 | $1.03 | $0.51 | −51 % |
| Email triage | 95.78% | 97.93% | ▲ 2.15 | $0.51 | $0.18 | −65 % |
| Personal data detection | 98.10% | 99.62% | ▲ 1.52 | $0.45 | $0.12 | −74 % |
| Contract clauses | 99.85% | 100.00% | ▲ 0.15 | $0.85 | $0.27 | −68 % |
| Document summaries | 99.56% | 100.00% | ▲ 0.44 | $0.69 | $0.21 | −69 % |
| Questions about a document | 100.00% | 100.00% | ±0 | $0.58 | $0.23 | −61 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference |
| Invoice extraction | 100.00% | 99.23% | ▼ 0.77 | $1.03 | $0.54 | −47 % |
| Email triage | 94.67% | 96.96% | ▲ 2.29 | $0.48 | $0.15 | −68 % |
| Personal data detection | 99.62% | 99.62% | ±0 | $0.45 | $0.12 | −74 % |
| Contract clauses | 98.00% | 100.00% | ▲ 2 | $0.79 | $0.27 | −66 % |
| Document summaries | 99.89% | 100.00% | ▲ 0.11 | $0.59 | $0.19 | −67 % |
| Questions about a document | 100.00% | 98.00% | ▼ 2 | $0.52 | $0.20 | −61 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference |
| Invoice extraction | 98.21% | 99.13% | ▲ 0.92 | $1.05 | $0.55 | −47 % |
| Email triage | 95.56% | 96.81% | ▲ 1.25 | $0.48 | $0.16 | −67 % |
| Personal data detection | 95.81% | 99.71% | ▲ 3.9 | $0.45 | $0.12 | −73 % |
| Contract clauses | 99.79% | 100.00% | ▲ 0.21 | $0.84 | $0.26 | −68 % |
| Document summaries | 95.33% | 97.00% | ▲ 1.67 | $0.56 | $0.19 | −66 % |
| Questions about a document | 100.00% | 99.67% | ▼ 0.33 | $0.51 | $0.19 | −62 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference | Gemini 3.5 Flash Lite | GPT-6 Luna | Difference |
| Invoice extraction | 98.97% | 99.08% | ▲ 0.11 | $1.07 | $0.56 | −48 % |
| Email triage | 96.15% | 97.26% | ▲ 1.11 | $0.49 | $0.16 | −67 % |
| Personal data detection | 94.95% | 99.62% | ▲ 4.67 | $0.46 | $0.12 | −73 % |
| Contract clauses | 99.95% | 99.33% | ▼ 0.62 | $0.85 | $0.27 | −68 % |
| Document summaries | 94.67% | 98.78% | ▲ 4.11 | $0.65 | $0.20 | −70 % |
| Questions about a document | 99.56% | 99.56% | ±0 | $0.53 | $0.20 | −62 % |
Difference: GPT-6 Luna vs. Gemini 3.5 Flash Lite; accuracy in percentage points, price in percent. The better value is in bold. Click a task to see the details.
Invoice extraction · English (US) (150 documents)
GPT-6 Luna is more accurate by 0.5 pp. GPT-6 Luna is 1.7× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.21% | 99.70% |
| Error-free documents | 92.00% | 96.67% |
| Per 1,000 documents | $0.87 | $0.51 |
| To fix / 1,000 | 80 | 33 |
| Speed (median) | 1.7 sec | 3.4 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Invoice extraction · Czech (150 documents)
Gemini 3.5 Flash Lite is more accurate by 0.2 pp. GPT-6 Luna is 2× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.49% | 99.28% |
| Error-free documents | 93.33% | 93.33% |
| Per 1,000 documents | $1.03 | $0.51 |
| To fix / 1,000 | 67 | 67 |
| Speed (median) | 1.8 sec | 3.2 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Invoice extraction · German (150 documents)
Gemini 3.5 Flash Lite is more accurate by 0.8 pp. GPT-6 Luna is 1.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.23% |
| Error-free documents | 100.00% | 91.33% |
| Per 1,000 documents | $1.03 | $0.54 |
| To fix / 1,000 | 0 | 87 |
| Speed (median) | 1.5 sec | 3.7 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Invoice extraction · Spanish (150 documents)
GPT-6 Luna is more accurate by 0.9 pp. GPT-6 Luna is 1.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 98.21% | 99.13% |
| Error-free documents | 78.00% | 88.67% |
| Per 1,000 documents | $1.05 | $0.55 |
| To fix / 1,000 | 220 | 113 |
| Speed (median) | 1.6 sec | 3.9 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Invoice extraction · French (150 documents)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 1.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 98.97% | 99.08% |
| Error-free documents | 88.00% | 88.00% |
| Per 1,000 documents | $1.07 | $0.56 |
| To fix / 1,000 | 120 | 120 |
| Speed (median) | 1.5 sec | 3.9 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Email triage · English (US) (150 documents)
GPT-6 Luna is more accurate by 1.2 pp. GPT-6 Luna is 3.1× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 96.30% | 97.48% |
| Error-free documents | 68.67% | 77.33% |
| Per 1,000 documents | $0.42 | $0.13 |
| To fix / 1,000 | 313 | 227 |
| Speed (median) | 0.7 sec | 2.5 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Email triage · Czech (150 documents)
GPT-6 Luna is more accurate by 2.1 pp. GPT-6 Luna is 2.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 95.78% | 97.93% |
| Error-free documents | 64.00% | 81.33% |
| Per 1,000 documents | $0.51 | $0.18 |
| To fix / 1,000 | 360 | 187 |
| Speed (median) | 0.7 sec | 2.6 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Email triage · German (150 documents)
GPT-6 Luna is more accurate by 2.3 pp. GPT-6 Luna is 3.1× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 94.67% | 96.96% |
| Error-free documents | 56.67% | 72.67% |
| Per 1,000 documents | $0.48 | $0.15 |
| To fix / 1,000 | 433 | 273 |
| Speed (median) | 0.8 sec | 2.8 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Email triage · Spanish (150 documents)
GPT-6 Luna is more accurate by 1.2 pp. GPT-6 Luna is 3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 95.56% | 96.81% |
| Error-free documents | 65.33% | 72.00% |
| Per 1,000 documents | $0.48 | $0.16 |
| To fix / 1,000 | 347 | 280 |
| Speed (median) | 0.8 sec | 3.1 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Email triage · French (150 documents)
GPT-6 Luna is more accurate by 1.1 pp. GPT-6 Luna is 3.1× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 96.15% | 97.26% |
| Error-free documents | 68.67% | 76.00% |
| Per 1,000 documents | $0.49 | $0.16 |
| To fix / 1,000 | 313 | 240 |
| Speed (median) | 0.8 sec | 2.8 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Personal data detection · English (US) (150 documents)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 3.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.52% | 99.62% |
| Error-free documents | 97.33% | 97.33% |
| Per 1,000 documents | $0.41 | $0.11 |
| To fix / 1,000 | 27 | 27 |
| Speed (median) | 0.8 sec | 1.8 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Personal data detection · Czech (150 documents)
GPT-6 Luna is more accurate by 1.5 pp. GPT-6 Luna is 3.8× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 98.10% | 99.62% |
| Error-free documents | 88.00% | 97.33% |
| Per 1,000 documents | $0.45 | $0.12 |
| To fix / 1,000 | 120 | 27 |
| Speed (median) | 0.8 sec | 1.8 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Personal data detection · German (150 documents)
Both are equally accurate. GPT-6 Luna is 3.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.62% | 99.62% |
| Error-free documents | 97.33% | 97.33% |
| Per 1,000 documents | $0.45 | $0.12 |
| To fix / 1,000 | 27 | 27 |
| Speed (median) | 0.8 sec | 2.2 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Personal data detection · Spanish (150 documents)
GPT-6 Luna is more accurate by 3.9 pp. GPT-6 Luna is 3.8× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 95.81% | 99.71% |
| Error-free documents | 73.33% | 98.00% |
| Per 1,000 documents | $0.45 | $0.12 |
| To fix / 1,000 | 267 | 20 |
| Speed (median) | 0.9 sec | 2.3 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Personal data detection · French (150 documents)
GPT-6 Luna is more accurate by 4.7 pp. GPT-6 Luna is 3.7× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 94.95% | 99.62% |
| Error-free documents | 67.33% | 97.33% |
| Per 1,000 documents | $0.46 | $0.12 |
| To fix / 1,000 | 327 | 27 |
| Speed (median) | 0.8 sec | 2.4 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Contract clauses · English (US) (150 documents)
Gemini 3.5 Flash Lite is more accurate by 0.2 pp. GPT-6 Luna is 3.4× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.95% | 99.79% |
| Error-free documents | 99.33% | 98.67% |
| Per 1,000 documents | $0.69 | $0.20 |
| To fix / 1,000 | 7 | 13 |
| Speed (median) | 1.0 sec | 2.7 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Contract clauses · Czech (150 documents)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 3.2× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.85% | 100.00% |
| Error-free documents | 98.00% | 100.00% |
| Per 1,000 documents | $0.85 | $0.27 |
| To fix / 1,000 | 20 | 0 |
| Speed (median) | 0.9 sec | 2.6 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Contract clauses · German (150 documents)
GPT-6 Luna is more accurate by 2 pp. GPT-6 Luna is 3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 98.00% | 100.00% |
| Error-free documents | 84.00% | 100.00% |
| Per 1,000 documents | $0.79 | $0.27 |
| To fix / 1,000 | 160 | 0 |
| Speed (median) | 1.0 sec | 3.0 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Contract clauses · Spanish (150 documents)
GPT-6 Luna is more accurate by 0.2 pp. GPT-6 Luna is 3.2× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.79% | 100.00% |
| Error-free documents | 98.00% | 100.00% |
| Per 1,000 documents | $0.84 | $0.26 |
| To fix / 1,000 | 20 | 0 |
| Speed (median) | 1.1 sec | 3.1 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Contract clauses · French (150 documents)
Gemini 3.5 Flash Lite is more accurate by 0.6 pp. GPT-6 Luna is 3.1× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.95% | 99.33% |
| Error-free documents | 99.33% | 91.33% |
| Per 1,000 documents | $0.85 | $0.27 |
| To fix / 1,000 | 7 | 87 |
| Speed (median) | 1.0 sec | 3.2 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Document summaries · English (US) (150 documents)
GPT-6 Luna is more accurate by 0.8 pp. GPT-6 Luna is 2.9× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.22% | 100.00% |
| Error-free documents | 95.33% | 100.00% |
| Per 1,000 documents | $0.46 | $0.16 |
| To fix / 1,000 | 47 | 0 |
| Speed (median) | 0.9 sec | 2.9 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Document summaries · Czech (150 documents)
GPT-6 Luna is more accurate by 0.4 pp. GPT-6 Luna is 3.3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.56% | 100.00% |
| Error-free documents | 97.33% | 100.00% |
| Per 1,000 documents | $0.69 | $0.21 |
| To fix / 1,000 | 27 | 0 |
| Speed (median) | 1.0 sec | 3.4 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Document summaries · German (150 documents)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.89% | 100.00% |
| Error-free documents | 99.33% | 100.00% |
| Per 1,000 documents | $0.59 | $0.19 |
| To fix / 1,000 | 7 | 0 |
| Speed (median) | 0.9 sec | 3.3 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Document summaries · Spanish (150 documents)
GPT-6 Luna is more accurate by 1.7 pp. GPT-6 Luna is 3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 95.33% | 97.00% |
| Error-free documents | 72.00% | 82.00% |
| Per 1,000 documents | $0.56 | $0.19 |
| To fix / 1,000 | 280 | 180 |
| Speed (median) | 0.9 sec | 3.0 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Document summaries · French (150 documents)
GPT-6 Luna is more accurate by 4.1 pp. GPT-6 Luna is 3.3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 94.67% | 98.78% |
| Error-free documents | 71.33% | 92.67% |
| Per 1,000 documents | $0.65 | $0.20 |
| To fix / 1,000 | 287 | 73 |
| Speed (median) | 0.9 sec | 3.2 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Questions about a document · English (US) (150 documents)
Gemini 3.5 Flash Lite is more accurate by 0.9 pp. GPT-6 Luna is 3× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.11% |
| Error-free documents | 100.00% | 96.00% |
| Per 1,000 documents | $0.46 | $0.15 |
| To fix / 1,000 | 0 | 40 |
| Speed (median) | 0.8 sec | 3.1 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Questions about a document · Czech (150 documents)
Both are equally accurate. GPT-6 Luna is 2.6× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 100.00% |
| Error-free documents | 100.00% | 100.00% |
| Per 1,000 documents | $0.58 | $0.23 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 0.8 sec | 3.1 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Questions about a document · German (150 documents)
Gemini 3.5 Flash Lite is more accurate by 2 pp. GPT-6 Luna is 2.6× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 98.00% |
| Error-free documents | 100.00% | 91.33% |
| Per 1,000 documents | $0.52 | $0.20 |
| To fix / 1,000 | 0 | 87 |
| Speed (median) | 0.7 sec | 3.2 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Questions about a document · Spanish (150 documents)
Gemini 3.5 Flash Lite is more accurate by 0.3 pp. GPT-6 Luna is 2.6× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.67% |
| Error-free documents | 100.00% | 98.67% |
| Per 1,000 documents | $0.51 | $0.19 |
| To fix / 1,000 | 0 | 13 |
| Speed (median) | 0.8 sec | 2.9 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Questions about a document · French (150 documents)
Both are equally accurate. GPT-6 Luna is 2.6× cheaper.
| Metric | Gemini 3.5 Flash Lite | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.56% | 99.56% |
| Error-free documents | 98.00% | 98.00% |
| Per 1,000 documents | $0.53 | $0.20 |
| To fix / 1,000 | 20 | 20 |
| Speed (median) | 0.8 sec | 3.0 sec |
| Valid answers | 100.00% | 100.00% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.