GPT-6 Luna vs Qwen3.8 Flash
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 19 of 30
Cheaper
GPT-6 Luna
cheaper in 30 of 30
Faster
GPT-6 Luna
3.0 sec vs 14.1 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 | GPT-6 Luna | Qwen3.8 Flash | Difference | GPT-6 Luna | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.70% | 82.36% | ▼ 17.34 | $0.51 | $1.15 | +127 % |
| Email triage | 97.48% | 97.48% | ±0 | $0.13 | $0.30 | +126 % |
| Personal data detection | 99.62% | 99.24% | ▼ 0.38 | $0.11 | $0.28 | +164 % |
| Contract clauses | 99.79% | 95.13% | ▼ 4.66 | $0.20 | $0.62 | +208 % |
| Document summaries | 100.00% | 99.89% | ▼ 0.11 | $0.16 | $0.75 | +368 % |
| Questions about a document | 99.11% | 99.78% | ▲ 0.67 | $0.15 | $0.44 | +188 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Luna | Qwen3.8 Flash | Difference | GPT-6 Luna | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.28% | 91.79% | ▼ 7.49 | $0.51 | $1.02 | +102 % |
| Email triage | 97.93% | 98.52% | ▲ 0.59 | $0.18 | $0.45 | +150 % |
| Personal data detection | 99.62% | 96.95% | ▼ 2.67 | $0.12 | $0.41 | +240 % |
| Contract clauses | 100.00% | 99.90% | ▼ 0.1 | $0.27 | $0.63 | +133 % |
| Document summaries | 100.00% | 95.44% | ▼ 4.56 | $0.21 | $1.48 | +597 % |
| Questions about a document | 100.00% | 94.22% | ▼ 5.78 | $0.23 | $0.71 | +214 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Luna | Qwen3.8 Flash | Difference | GPT-6 Luna | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.23% | 93.49% | ▼ 5.74 | $0.54 | $1.12 | +105 % |
| Email triage | 96.96% | 97.85% | ▲ 0.89 | $0.15 | $0.39 | +157 % |
| Personal data detection | 99.62% | 99.62% | ±0 | $0.12 | $0.36 | +206 % |
| Contract clauses | 100.00% | 98.00% | ▼ 2 | $0.27 | $0.78 | +193 % |
| Document summaries | 100.00% | 95.33% | ▼ 4.67 | $0.19 | $1.43 | +636 % |
| Questions about a document | 98.00% | 98.89% | ▲ 0.89 | $0.20 | $0.46 | +129 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Luna | Qwen3.8 Flash | Difference | GPT-6 Luna | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.13% | 88.87% | ▼ 10.26 | $0.55 | $1.38 | +150 % |
| Email triage | 96.81% | 96.74% | ▼ 0.07 | $0.16 | $0.46 | +192 % |
| Personal data detection | 99.71% | 99.81% | ▲ 0.1 | $0.12 | $0.38 | +219 % |
| Contract clauses | 100.00% | 100.00% | ±0 | $0.26 | $0.60 | +125 % |
| Document summaries | 97.00% | 94.44% | ▼ 2.56 | $0.19 | $1.22 | +548 % |
| Questions about a document | 99.67% | 99.67% | ±0 | $0.19 | $0.53 | +170 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Luna | Qwen3.8 Flash | Difference | GPT-6 Luna | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.08% | 90.46% | ▼ 8.62 | $0.56 | $1.28 | +129 % |
| Email triage | 97.26% | 96.44% | ▼ 0.82 | $0.16 | $0.45 | +183 % |
| Personal data detection | 99.62% | 99.71% | ▲ 0.09 | $0.12 | $0.40 | +219 % |
| Contract clauses | 99.33% | 99.49% | ▲ 0.16 | $0.27 | $0.66 | +141 % |
| Document summaries | 98.78% | 96.33% | ▼ 2.45 | $0.20 | $1.46 | +642 % |
| Questions about a document | 99.56% | 98.78% | ▼ 0.78 | $0.20 | $0.57 | +178 % |
Difference: Qwen3.8 Flash vs. GPT-6 Luna; 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 17.3 pp. GPT-6 Luna is 2.3× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.70% | 82.36% |
| Error-free documents | 96.67% | 57.33% |
| Per 1,000 documents | $0.51 | $1.15 |
| To fix / 1,000 | 33 | 427 |
| Speed (median) | 3.4 sec | 17.4 sec |
| Valid answers | 100.00% | 90.67% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Invoice extraction · Czech (150 documents)
GPT-6 Luna is more accurate by 7.5 pp. GPT-6 Luna is 2× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.28% | 91.79% |
| Error-free documents | 93.33% | 76.67% |
| Per 1,000 documents | $0.51 | $1.02 |
| To fix / 1,000 | 67 | 233 |
| Speed (median) | 3.2 sec | 17.0 sec |
| Valid answers | 100.00% | 96.67% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Invoice extraction · German (150 documents)
GPT-6 Luna is more accurate by 5.7 pp. GPT-6 Luna is 2.1× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.23% | 93.49% |
| Error-free documents | 91.33% | 87.33% |
| Per 1,000 documents | $0.54 | $1.12 |
| To fix / 1,000 | 87 | 127 |
| Speed (median) | 3.7 sec | 19.0 sec |
| Valid answers | 100.00% | 94.00% |
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 10.3 pp. GPT-6 Luna is 2.5× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.13% | 88.87% |
| Error-free documents | 88.67% | 74.67% |
| Per 1,000 documents | $0.55 | $1.38 |
| To fix / 1,000 | 113 | 253 |
| Speed (median) | 3.9 sec | 22.0 sec |
| Valid answers | 100.00% | 90.67% |
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 8.6 pp. GPT-6 Luna is 2.3× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.08% | 90.46% |
| Error-free documents | 88.00% | 83.33% |
| Per 1,000 documents | $0.56 | $1.28 |
| To fix / 1,000 | 120 | 167 |
| Speed (median) | 3.9 sec | 19.4 sec |
| Valid answers | 100.00% | 94.67% |
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)
Both are equally accurate. GPT-6 Luna is 2.3× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 97.48% | 97.48% |
| Error-free documents | 77.33% | 78.67% |
| Per 1,000 documents | $0.13 | $0.30 |
| To fix / 1,000 | 227 | 213 |
| Speed (median) | 2.5 sec | 8.9 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 · Czech (150 documents)
Qwen3.8 Flash is more accurate by 0.6 pp. GPT-6 Luna is 2.5× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 97.93% | 98.52% |
| Error-free documents | 81.33% | 86.67% |
| Per 1,000 documents | $0.18 | $0.45 |
| To fix / 1,000 | 187 | 133 |
| Speed (median) | 2.6 sec | 13.0 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.
Email triage · German (150 documents)
Qwen3.8 Flash is more accurate by 0.9 pp. GPT-6 Luna is 2.6× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 96.96% | 97.85% |
| Error-free documents | 72.67% | 81.33% |
| Per 1,000 documents | $0.15 | $0.39 |
| To fix / 1,000 | 273 | 187 |
| Speed (median) | 2.8 sec | 12.1 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 · Spanish (150 documents)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 2.9× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 96.81% | 96.74% |
| Error-free documents | 72.00% | 72.00% |
| Per 1,000 documents | $0.16 | $0.46 |
| To fix / 1,000 | 280 | 280 |
| Speed (median) | 3.1 sec | 11.2 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.
Email triage · French (150 documents)
GPT-6 Luna is more accurate by 0.8 pp. GPT-6 Luna is 2.8× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 97.26% | 96.44% |
| Error-free documents | 76.00% | 76.00% |
| Per 1,000 documents | $0.16 | $0.45 |
| To fix / 1,000 | 240 | 240 |
| Speed (median) | 2.8 sec | 13.2 sec |
| Valid answers | 100.00% | 99.33% |
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 0.4 pp. GPT-6 Luna is 2.6× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.62% | 99.24% |
| Error-free documents | 97.33% | 98.67% |
| Per 1,000 documents | $0.11 | $0.28 |
| To fix / 1,000 | 27 | 13 |
| Speed (median) | 1.8 sec | 7.5 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.
Personal data detection · Czech (150 documents)
GPT-6 Luna is more accurate by 2.7 pp. GPT-6 Luna is 3.4× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.62% | 96.95% |
| Error-free documents | 97.33% | 88.67% |
| Per 1,000 documents | $0.12 | $0.41 |
| To fix / 1,000 | 27 | 113 |
| Speed (median) | 1.8 sec | 9.6 sec |
| Valid answers | 100.00% | 99.33% |
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 3.1× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.62% | 99.62% |
| Error-free documents | 97.33% | 99.33% |
| Per 1,000 documents | $0.12 | $0.36 |
| To fix / 1,000 | 27 | 7 |
| Speed (median) | 2.2 sec | 12.2 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)
Qwen3.8 Flash is more accurate by 0.1 pp. GPT-6 Luna is 3.2× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.71% | 99.81% |
| Error-free documents | 98.00% | 98.67% |
| Per 1,000 documents | $0.12 | $0.38 |
| To fix / 1,000 | 20 | 13 |
| Speed (median) | 2.3 sec | 10.2 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 · French (150 documents)
Qwen3.8 Flash is more accurate by 0.1 pp. GPT-6 Luna is 3.2× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.62% | 99.71% |
| Error-free documents | 97.33% | 98.00% |
| Per 1,000 documents | $0.12 | $0.40 |
| To fix / 1,000 | 27 | 20 |
| Speed (median) | 2.4 sec | 10.9 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.
Contract clauses · English (US) (150 documents)
GPT-6 Luna is more accurate by 4.7 pp. GPT-6 Luna is 3.1× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.79% | 95.13% |
| Error-free documents | 98.67% | 69.33% |
| Per 1,000 documents | $0.20 | $0.62 |
| To fix / 1,000 | 13 | 307 |
| Speed (median) | 2.7 sec | 14.1 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 · Czech (150 documents)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 2.3× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 99.90% |
| Error-free documents | 100.00% | 98.67% |
| Per 1,000 documents | $0.27 | $0.63 |
| To fix / 1,000 | 0 | 13 |
| Speed (median) | 2.6 sec | 13.6 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 2.9× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 98.00% |
| Error-free documents | 100.00% | 87.33% |
| Per 1,000 documents | $0.27 | $0.78 |
| To fix / 1,000 | 0 | 127 |
| Speed (median) | 3.0 sec | 19.7 sec |
| Valid answers | 100.00% | 99.33% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.
Contract clauses · Spanish (150 documents)
Both are equally accurate. GPT-6 Luna is 2.3× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 100.00% |
| Error-free documents | 100.00% | 100.00% |
| Per 1,000 documents | $0.26 | $0.60 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 3.1 sec | 14.2 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.
Contract clauses · French (150 documents)
Qwen3.8 Flash is more accurate by 0.2 pp. GPT-6 Luna is 2.4× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.33% | 99.49% |
| Error-free documents | 91.33% | 93.33% |
| Per 1,000 documents | $0.27 | $0.66 |
| To fix / 1,000 | 87 | 67 |
| Speed (median) | 3.2 sec | 16.2 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 0.1 pp. GPT-6 Luna is 4.7× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 99.89% |
| Error-free documents | 100.00% | 99.33% |
| Per 1,000 documents | $0.16 | $0.75 |
| To fix / 1,000 | 0 | 7 |
| Speed (median) | 2.9 sec | 17.8 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.
Document summaries · Czech (150 documents)
GPT-6 Luna is more accurate by 4.6 pp. GPT-6 Luna is 7× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 95.44% |
| Error-free documents | 100.00% | 93.33% |
| Per 1,000 documents | $0.21 | $1.48 |
| To fix / 1,000 | 0 | 67 |
| Speed (median) | 3.4 sec | 34.2 sec |
| Valid answers | 100.00% | 96.00% |
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 4.7 pp. GPT-6 Luna is 7.4× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 95.33% |
| Error-free documents | 100.00% | 95.33% |
| Per 1,000 documents | $0.19 | $1.43 |
| To fix / 1,000 | 0 | 47 |
| Speed (median) | 3.3 sec | 29.5 sec |
| Valid answers | 100.00% | 95.33% |
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 2.6 pp. GPT-6 Luna is 6.5× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 97.00% | 94.44% |
| Error-free documents | 82.00% | 93.33% |
| Per 1,000 documents | $0.19 | $1.22 |
| To fix / 1,000 | 180 | 67 |
| Speed (median) | 3.0 sec | 23.9 sec |
| Valid answers | 100.00% | 94.67% |
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 2.4 pp. GPT-6 Luna is 7.4× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 98.78% | 96.33% |
| Error-free documents | 92.67% | 94.67% |
| Per 1,000 documents | $0.20 | $1.46 |
| To fix / 1,000 | 73 | 53 |
| Speed (median) | 3.2 sec | 28.1 sec |
| Valid answers | 100.00% | 96.67% |
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)
Qwen3.8 Flash is more accurate by 0.7 pp. GPT-6 Luna is 2.9× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.11% | 99.78% |
| Error-free documents | 96.00% | 98.67% |
| Per 1,000 documents | $0.15 | $0.44 |
| To fix / 1,000 | 40 | 13 |
| Speed (median) | 3.1 sec | 8.7 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.
Questions about a document · Czech (150 documents)
GPT-6 Luna is more accurate by 5.8 pp. GPT-6 Luna is 3.1× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 94.22% |
| Error-free documents | 100.00% | 84.67% |
| Per 1,000 documents | $0.23 | $0.71 |
| To fix / 1,000 | 0 | 153 |
| Speed (median) | 3.1 sec | 14.3 sec |
| Valid answers | 100.00% | 98.67% |
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)
Qwen3.8 Flash is more accurate by 0.9 pp. GPT-6 Luna is 2.3× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 98.00% | 98.89% |
| Error-free documents | 91.33% | 96.67% |
| Per 1,000 documents | $0.20 | $0.46 |
| To fix / 1,000 | 87 | 33 |
| Speed (median) | 3.2 sec | 8.6 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.
Questions about a document · Spanish (150 documents)
Both are equally accurate. GPT-6 Luna is 2.7× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.67% | 99.67% |
| Error-free documents | 98.67% | 98.67% |
| Per 1,000 documents | $0.19 | $0.53 |
| To fix / 1,000 | 13 | 13 |
| Speed (median) | 2.9 sec | 10.3 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.
Questions about a document · French (150 documents)
GPT-6 Luna is more accurate by 0.8 pp. GPT-6 Luna is 2.8× cheaper.
| Metric | GPT-6 Luna | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.56% | 98.78% |
| Error-free documents | 98.00% | 97.33% |
| Per 1,000 documents | $0.20 | $0.57 |
| To fix / 1,000 | 20 | 27 |
| Speed (median) | 3.0 sec | 10.4 sec |
| Valid answers | 100.00% | 99.33% |
Free with login
Accuracy field by field and the documents where the models differ.
Just your email, no password.