GPT-6 Sol 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 Sol is more accurate; Qwen3.8 Flash is cheaper in 30 of 30 combinations.
More accurate
GPT-6 Sol
wins 26 of 30
Cheaper
Qwen3.8 Flash
cheaper in 30 of 30
Faster
GPT-6 Sol
2.9 sec vs 14.1 sec
Fewer fixes
GPT-6 Sol
13 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 Sol | Qwen3.8 Flash | Difference | GPT-6 Sol | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.94% | 82.36% | ▼ 17.58 | $9.65 | $1.15 | −88 % |
| Email triage | 99.04% | 97.48% | ▼ 1.56 | $2.44 | $0.30 | −88 % |
| Personal data detection | 99.71% | 99.24% | ▼ 0.47 | $2.09 | $0.28 | −87 % |
| Contract clauses | 100.00% | 95.13% | ▼ 4.87 | $3.33 | $0.62 | −81 % |
| Document summaries | 99.56% | 99.89% | ▲ 0.33 | $3.85 | $0.75 | −81 % |
| Questions about a document | 100.00% | 99.78% | ▼ 0.22 | $2.36 | $0.44 | −81 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Sol | Qwen3.8 Flash | Difference | GPT-6 Sol | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.90% | 91.79% | ▼ 8.11 | $10.12 | $1.02 | −90 % |
| Email triage | 98.96% | 98.52% | ▼ 0.44 | $3.19 | $0.45 | −86 % |
| Personal data detection | 99.71% | 96.95% | ▼ 2.76 | $2.22 | $0.41 | −82 % |
| Contract clauses | 100.00% | 99.90% | ▼ 0.1 | $4.62 | $0.63 | −86 % |
| Document summaries | 100.00% | 95.44% | ▼ 4.56 | $4.72 | $1.48 | −69 % |
| Questions about a document | 100.00% | 94.22% | ▼ 5.78 | $3.95 | $0.71 | −82 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Sol | Qwen3.8 Flash | Difference | GPT-6 Sol | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.95% | 93.49% | ▼ 6.46 | $10.65 | $1.12 | −89 % |
| Email triage | 98.22% | 97.85% | ▼ 0.37 | $2.73 | $0.39 | −86 % |
| Personal data detection | 99.71% | 99.62% | ▼ 0.09 | $2.13 | $0.36 | −83 % |
| Contract clauses | 100.00% | 98.00% | ▼ 2 | $4.36 | $0.78 | −82 % |
| Document summaries | 100.00% | 95.33% | ▼ 4.67 | $4.11 | $1.43 | −65 % |
| Questions about a document | 100.00% | 98.89% | ▼ 1.11 | $3.23 | $0.46 | −86 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Sol | Qwen3.8 Flash | Difference | GPT-6 Sol | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.23% | 88.87% | ▼ 10.36 | $10.79 | $1.38 | −87 % |
| Email triage | 98.44% | 96.74% | ▼ 1.7 | $2.77 | $0.46 | −83 % |
| Personal data detection | 99.81% | 99.81% | ±0 | $2.21 | $0.38 | −83 % |
| Contract clauses | 100.00% | 100.00% | ±0 | $4.48 | $0.60 | −87 % |
| Document summaries | 97.89% | 94.44% | ▼ 3.45 | $5.02 | $1.22 | −76 % |
| Questions about a document | 100.00% | 99.67% | ▼ 0.33 | $3.36 | $0.53 | −84 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | GPT-6 Sol | Qwen3.8 Flash | Difference | GPT-6 Sol | Qwen3.8 Flash | Difference |
| Invoice extraction | 99.90% | 90.46% | ▼ 9.44 | $11.00 | $1.28 | −88 % |
| Email triage | 98.59% | 96.44% | ▼ 2.15 | $2.81 | $0.45 | −84 % |
| Personal data detection | 99.71% | 99.71% | ±0 | $2.23 | $0.40 | −82 % |
| Contract clauses | 99.59% | 99.49% | ▼ 0.1 | $4.49 | $0.66 | −85 % |
| Document summaries | 97.22% | 96.33% | ▼ 0.89 | $4.58 | $1.46 | −68 % |
| Questions about a document | 100.00% | 98.78% | ▼ 1.22 | $3.43 | $0.57 | −83 % |
Difference: Qwen3.8 Flash vs. GPT-6 Sol; 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 Sol is more accurate by 17.6 pp. Qwen3.8 Flash is 8.4× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.94% | 82.36% |
| Error-free documents | 99.33% | 57.33% |
| Per 1,000 documents | $9.65 | $1.15 |
| To fix / 1,000 | 7 | 427 |
| Speed (median) | 4.1 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 Sol is more accurate by 8.1 pp. Qwen3.8 Flash is 9.9× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.90% | 91.79% |
| Error-free documents | 98.67% | 76.67% |
| Per 1,000 documents | $10.12 | $1.02 |
| To fix / 1,000 | 13 | 233 |
| Speed (median) | 3.9 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 Sol is more accurate by 6.5 pp. Qwen3.8 Flash is 9.5× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.95% | 93.49% |
| Error-free documents | 99.33% | 87.33% |
| Per 1,000 documents | $10.65 | $1.12 |
| To fix / 1,000 | 7 | 127 |
| Speed (median) | 4.9 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 Sol is more accurate by 10.4 pp. Qwen3.8 Flash is 7.8× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.23% | 88.87% |
| Error-free documents | 90.00% | 74.67% |
| Per 1,000 documents | $10.79 | $1.38 |
| To fix / 1,000 | 100 | 253 |
| Speed (median) | 5.9 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 Sol is more accurate by 9.4 pp. Qwen3.8 Flash is 8.6× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.90% | 90.46% |
| Error-free documents | 98.67% | 83.33% |
| Per 1,000 documents | $11.00 | $1.28 |
| To fix / 1,000 | 13 | 167 |
| Speed (median) | 6.0 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)
GPT-6 Sol is more accurate by 1.6 pp. Qwen3.8 Flash is 8× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.04% | 97.48% |
| Error-free documents | 92.00% | 78.67% |
| Per 1,000 documents | $2.44 | $0.30 |
| To fix / 1,000 | 80 | 213 |
| Speed (median) | 2.8 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)
GPT-6 Sol is more accurate by 0.4 pp. Qwen3.8 Flash is 7.2× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 98.96% | 98.52% |
| Error-free documents | 90.67% | 86.67% |
| Per 1,000 documents | $3.19 | $0.45 |
| To fix / 1,000 | 93 | 133 |
| Speed (median) | 2.8 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)
GPT-6 Sol is more accurate by 0.4 pp. Qwen3.8 Flash is 7× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 98.22% | 97.85% |
| Error-free documents | 84.00% | 81.33% |
| Per 1,000 documents | $2.73 | $0.39 |
| To fix / 1,000 | 160 | 187 |
| Speed (median) | 2.9 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 Sol is more accurate by 1.7 pp. Qwen3.8 Flash is 6× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 98.44% | 96.74% |
| Error-free documents | 86.00% | 72.00% |
| Per 1,000 documents | $2.77 | $0.46 |
| To fix / 1,000 | 140 | 280 |
| Speed (median) | 3.3 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 Sol is more accurate by 2.1 pp. Qwen3.8 Flash is 6.2× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 98.59% | 96.44% |
| Error-free documents | 87.33% | 76.00% |
| Per 1,000 documents | $2.81 | $0.45 |
| To fix / 1,000 | 127 | 240 |
| Speed (median) | 3.2 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 Sol is more accurate by 0.5 pp. Qwen3.8 Flash is 7.5× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.71% | 99.24% |
| Error-free documents | 98.00% | 98.67% |
| Per 1,000 documents | $2.09 | $0.28 |
| To fix / 1,000 | 20 | 13 |
| Speed (median) | 2.6 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 Sol is more accurate by 2.8 pp. Qwen3.8 Flash is 5.5× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.71% | 96.95% |
| Error-free documents | 98.00% | 88.67% |
| Per 1,000 documents | $2.22 | $0.41 |
| To fix / 1,000 | 20 | 113 |
| Speed (median) | 2.5 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)
GPT-6 Sol is more accurate by 0.1 pp. Qwen3.8 Flash is 5.9× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.71% | 99.62% |
| Error-free documents | 98.00% | 99.33% |
| Per 1,000 documents | $2.13 | $0.36 |
| To fix / 1,000 | 20 | 7 |
| Speed (median) | 2.6 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)
Both are equally accurate. Qwen3.8 Flash is 5.9× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.81% | 99.81% |
| Error-free documents | 98.67% | 98.67% |
| Per 1,000 documents | $2.21 | $0.38 |
| To fix / 1,000 | 13 | 13 |
| Speed (median) | 2.9 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)
Both are equally accurate. Qwen3.8 Flash is 5.6× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.71% | 99.71% |
| Error-free documents | 98.00% | 98.00% |
| Per 1,000 documents | $2.23 | $0.40 |
| To fix / 1,000 | 20 | 20 |
| Speed (median) | 2.9 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 Sol is more accurate by 4.9 pp. Qwen3.8 Flash is 5.4× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 95.13% |
| Error-free documents | 100.00% | 69.33% |
| Per 1,000 documents | $3.33 | $0.62 |
| To fix / 1,000 | 0 | 307 |
| Speed (median) | 2.5 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 Sol is more accurate by 0.1 pp. Qwen3.8 Flash is 7.3× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 99.90% |
| Error-free documents | 100.00% | 98.67% |
| Per 1,000 documents | $4.62 | $0.63 |
| To fix / 1,000 | 0 | 13 |
| Speed (median) | 2.0 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 Sol is more accurate by 2 pp. Qwen3.8 Flash is 5.6× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 98.00% |
| Error-free documents | 100.00% | 87.33% |
| Per 1,000 documents | $4.36 | $0.78 |
| To fix / 1,000 | 0 | 127 |
| Speed (median) | 2.9 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. Qwen3.8 Flash is 7.5× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 100.00% |
| Error-free documents | 100.00% | 100.00% |
| Per 1,000 documents | $4.48 | $0.60 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 2.8 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)
GPT-6 Sol is more accurate by 0.1 pp. Qwen3.8 Flash is 6.8× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.59% | 99.49% |
| Error-free documents | 94.67% | 93.33% |
| Per 1,000 documents | $4.49 | $0.66 |
| To fix / 1,000 | 53 | 67 |
| Speed (median) | 2.6 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)
Qwen3.8 Flash is more accurate by 0.3 pp. Qwen3.8 Flash is 5.2× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 99.56% | 99.89% |
| Error-free documents | 97.33% | 99.33% |
| Per 1,000 documents | $3.85 | $0.75 |
| To fix / 1,000 | 27 | 7 |
| Speed (median) | 8.1 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 Sol is more accurate by 4.6 pp. Qwen3.8 Flash is 3.2× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 95.44% |
| Error-free documents | 100.00% | 93.33% |
| Per 1,000 documents | $4.72 | $1.48 |
| To fix / 1,000 | 0 | 67 |
| Speed (median) | 7.6 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 Sol is more accurate by 4.7 pp. Qwen3.8 Flash is 2.9× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 95.33% |
| Error-free documents | 100.00% | 95.33% |
| Per 1,000 documents | $4.11 | $1.43 |
| To fix / 1,000 | 0 | 47 |
| Speed (median) | 7.1 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 Sol is more accurate by 3.4 pp. Qwen3.8 Flash is 4.1× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 97.89% | 94.44% |
| Error-free documents | 87.33% | 93.33% |
| Per 1,000 documents | $5.02 | $1.22 |
| To fix / 1,000 | 127 | 67 |
| Speed (median) | 7.0 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 Sol is more accurate by 0.9 pp. Qwen3.8 Flash is 3.1× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 97.22% | 96.33% |
| Error-free documents | 83.33% | 94.67% |
| Per 1,000 documents | $4.58 | $1.46 |
| To fix / 1,000 | 167 | 53 |
| Speed (median) | 6.1 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)
GPT-6 Sol is more accurate by 0.2 pp. Qwen3.8 Flash is 5.4× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 99.78% |
| Error-free documents | 100.00% | 98.67% |
| Per 1,000 documents | $2.36 | $0.44 |
| To fix / 1,000 | 0 | 13 |
| Speed (median) | 2.3 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 Sol is more accurate by 5.8 pp. Qwen3.8 Flash is 5.5× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 94.22% |
| Error-free documents | 100.00% | 84.67% |
| Per 1,000 documents | $3.95 | $0.71 |
| To fix / 1,000 | 0 | 153 |
| Speed (median) | 2.3 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)
GPT-6 Sol is more accurate by 1.1 pp. Qwen3.8 Flash is 7× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 98.89% |
| Error-free documents | 100.00% | 96.67% |
| Per 1,000 documents | $3.23 | $0.46 |
| To fix / 1,000 | 0 | 33 |
| Speed (median) | 2.2 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)
GPT-6 Sol is more accurate by 0.3 pp. Qwen3.8 Flash is 6.4× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 99.67% |
| Error-free documents | 100.00% | 98.67% |
| Per 1,000 documents | $3.36 | $0.53 |
| To fix / 1,000 | 0 | 13 |
| Speed (median) | 2.5 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 Sol is more accurate by 1.2 pp. Qwen3.8 Flash is 6× cheaper.
| Metric | GPT-6 Sol | Qwen3.8 Flash |
|---|---|---|
| Correct fields | 100.00% | 98.78% |
| Error-free documents | 100.00% | 97.33% |
| Per 1,000 documents | $3.43 | $0.57 |
| To fix / 1,000 | 0 | 27 |
| Speed (median) | 2.5 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.