Claude Haiku 5.5 vs GPT-6 Luna
Same documents, same prompt: where each model gets it right, what it really costs, and how fast it responds.
Verdict
Claude Haiku 5.5 is more accurate; GPT-6 Luna is cheaper in 30 of 30 combinations.
More accurate
Claude Haiku 5.5
wins 15 of 30
Cheaper
GPT-6 Luna
cheaper in 30 of 30
Faster
Claude Haiku 5.5
2.1 sec vs 3.0 sec
Fewer fixes
Tie
33 vs 33 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 | Claude Haiku 5.5 | GPT-6 Luna | Difference | Claude Haiku 5.5 | GPT-6 Luna | Difference |
| Invoice extraction | 100.00% | 99.70% | ▼ 0.3 | $0.61 | $0.51 | −16 % |
| Email triage | 98.15% | 97.48% | ▼ 0.67 | $0.23 | $0.13 | −43 % |
| Personal data detection | 99.81% | 99.62% | ▼ 0.19 | $0.19 | $0.11 | −44 % |
| Contract clauses | 99.69% | 99.79% | ▲ 0.1 | $0.35 | $0.20 | −43 % |
| Document summaries | 98.33% | 100.00% | ▲ 1.67 | $0.34 | $0.16 | −53 % |
| Questions about a document | 100.00% | 99.11% | ▼ 0.89 | $0.23 | $0.15 | −34 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Claude Haiku 5.5 | GPT-6 Luna | Difference | Claude Haiku 5.5 | GPT-6 Luna | Difference |
| Invoice extraction | 99.90% | 99.28% | ▼ 0.62 | $0.62 | $0.51 | −19 % |
| Email triage | 97.63% | 97.93% | ▲ 0.3 | $0.29 | $0.18 | −38 % |
| Personal data detection | 99.43% | 99.62% | ▲ 0.19 | $0.21 | $0.12 | −43 % |
| Contract clauses | 100.00% | 100.00% | ±0 | $0.40 | $0.27 | −32 % |
| Document summaries | 95.89% | 100.00% | ▲ 4.11 | $0.32 | $0.21 | −35 % |
| Questions about a document | 99.78% | 100.00% | ▲ 0.22 | $0.30 | $0.23 | −25 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Claude Haiku 5.5 | GPT-6 Luna | Difference | Claude Haiku 5.5 | GPT-6 Luna | Difference |
| Invoice extraction | 99.85% | 99.23% | ▼ 0.62 | $0.71 | $0.54 | −23 % |
| Email triage | 97.11% | 96.96% | ▼ 0.15 | $0.32 | $0.15 | −52 % |
| Personal data detection | 99.62% | 99.62% | ±0 | $0.23 | $0.12 | −49 % |
| Contract clauses | 100.00% | 100.00% | ±0 | $0.48 | $0.27 | −45 % |
| Document summaries | 99.44% | 100.00% | ▲ 0.56 | $0.46 | $0.19 | −58 % |
| Questions about a document | 100.00% | 98.00% | ▼ 2 | $0.32 | $0.20 | −38 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Claude Haiku 5.5 | GPT-6 Luna | Difference | Claude Haiku 5.5 | GPT-6 Luna | Difference |
| Invoice extraction | 99.44% | 99.13% | ▼ 0.31 | $0.73 | $0.55 | −24 % |
| Email triage | 97.70% | 96.81% | ▼ 0.89 | $0.30 | $0.16 | −47 % |
| Personal data detection | 100.00% | 99.71% | ▼ 0.29 | $0.23 | $0.12 | −48 % |
| Contract clauses | 99.95% | 100.00% | ▲ 0.05 | $0.42 | $0.26 | −37 % |
| Document summaries | 94.11% | 97.00% | ▲ 2.89 | $0.61 | $0.19 | −69 % |
| Questions about a document | 100.00% | 99.67% | ▼ 0.33 | $0.29 | $0.19 | −32 % |
| Accuracy (correct fields) | Per 1,000 documents | |||||
|---|---|---|---|---|---|---|
| Task · documents | Claude Haiku 5.5 | GPT-6 Luna | Difference | Claude Haiku 5.5 | GPT-6 Luna | Difference |
| Invoice extraction | 99.85% | 99.08% | ▼ 0.77 | $0.77 | $0.56 | −28 % |
| Email triage | 97.41% | 97.26% | ▼ 0.15 | $0.31 | $0.16 | −49 % |
| Personal data detection | 99.33% | 99.62% | ▲ 0.29 | $0.23 | $0.12 | −46 % |
| Contract clauses | 99.38% | 99.33% | ▼ 0.05 | $0.45 | $0.27 | −40 % |
| Document summaries | 92.78% | 98.78% | ▲ 6 | $0.66 | $0.20 | −70 % |
| Questions about a document | 99.22% | 99.56% | ▲ 0.34 | $0.28 | $0.20 | −28 % |
Difference: GPT-6 Luna vs. Claude Haiku 5.5; 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)
Claude Haiku 5.5 is more accurate by 0.3 pp. GPT-6 Luna is 1.2× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.70% |
| Error-free documents | 100.00% | 96.67% |
| Per 1,000 documents | $0.61 | $0.51 |
| To fix / 1,000 | 0 | 33 |
| Speed (median) | 3.4 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)
Claude Haiku 5.5 is more accurate by 0.6 pp. GPT-6 Luna is 1.2× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.90% | 99.28% |
| Error-free documents | 98.67% | 93.33% |
| Per 1,000 documents | $0.62 | $0.51 |
| To fix / 1,000 | 13 | 67 |
| Speed (median) | 3.4 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)
Claude Haiku 5.5 is more accurate by 0.6 pp. GPT-6 Luna is 1.3× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.85% | 99.23% |
| Error-free documents | 98.00% | 91.33% |
| Per 1,000 documents | $0.71 | $0.54 |
| To fix / 1,000 | 20 | 87 |
| Speed (median) | 3.9 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)
Claude Haiku 5.5 is more accurate by 0.3 pp. GPT-6 Luna is 1.3× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.44% | 99.13% |
| Error-free documents | 92.67% | 88.67% |
| Per 1,000 documents | $0.73 | $0.55 |
| To fix / 1,000 | 73 | 113 |
| Speed (median) | 4.4 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)
Claude Haiku 5.5 is more accurate by 0.8 pp. GPT-6 Luna is 1.4× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.85% | 99.08% |
| Error-free documents | 98.00% | 88.00% |
| Per 1,000 documents | $0.77 | $0.56 |
| To fix / 1,000 | 20 | 120 |
| Speed (median) | 4.6 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.
Email triage · English (US) (150 documents)
Claude Haiku 5.5 is more accurate by 0.7 pp. GPT-6 Luna is 1.7× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 98.15% | 97.48% |
| Error-free documents | 84.00% | 77.33% |
| Per 1,000 documents | $0.23 | $0.13 |
| To fix / 1,000 | 160 | 227 |
| Speed (median) | 1.7 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 0.3 pp. GPT-6 Luna is 1.6× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 97.63% | 97.93% |
| Error-free documents | 80.00% | 81.33% |
| Per 1,000 documents | $0.29 | $0.18 |
| To fix / 1,000 | 200 | 187 |
| Speed (median) | 1.9 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.
Email triage · German (150 documents)
Claude Haiku 5.5 is more accurate by 0.1 pp. GPT-6 Luna is 2.1× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 97.11% | 96.96% |
| Error-free documents | 74.67% | 72.67% |
| Per 1,000 documents | $0.32 | $0.15 |
| To fix / 1,000 | 253 | 273 |
| Speed (median) | 2.1 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)
Claude Haiku 5.5 is more accurate by 0.9 pp. GPT-6 Luna is 1.9× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 97.70% | 96.81% |
| Error-free documents | 80.00% | 72.00% |
| Per 1,000 documents | $0.30 | $0.16 |
| To fix / 1,000 | 200 | 280 |
| Speed (median) | 2.1 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.
Email triage · French (150 documents)
Claude Haiku 5.5 is more accurate by 0.1 pp. GPT-6 Luna is 2× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 97.41% | 97.26% |
| Error-free documents | 76.67% | 76.00% |
| Per 1,000 documents | $0.31 | $0.16 |
| To fix / 1,000 | 233 | 240 |
| Speed (median) | 2.2 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)
Claude Haiku 5.5 is more accurate by 0.2 pp. GPT-6 Luna is 1.8× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.81% | 99.62% |
| Error-free documents | 98.67% | 97.33% |
| Per 1,000 documents | $0.19 | $0.11 |
| To fix / 1,000 | 13 | 27 |
| Speed (median) | 1.7 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 0.2 pp. GPT-6 Luna is 1.8× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.43% | 99.62% |
| Error-free documents | 96.00% | 97.33% |
| Per 1,000 documents | $0.21 | $0.12 |
| To fix / 1,000 | 40 | 27 |
| Speed (median) | 1.8 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 2× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.62% | 99.62% |
| Error-free documents | 97.33% | 97.33% |
| Per 1,000 documents | $0.23 | $0.12 |
| To fix / 1,000 | 27 | 27 |
| Speed (median) | 1.8 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)
Claude Haiku 5.5 is more accurate by 0.3 pp. GPT-6 Luna is 1.9× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.71% |
| Error-free documents | 100.00% | 98.00% |
| Per 1,000 documents | $0.23 | $0.12 |
| To fix / 1,000 | 0 | 20 |
| Speed (median) | 1.9 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 0.3 pp. GPT-6 Luna is 1.9× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.33% | 99.62% |
| Error-free documents | 95.33% | 97.33% |
| Per 1,000 documents | $0.23 | $0.12 |
| To fix / 1,000 | 47 | 27 |
| Speed (median) | 1.8 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)
GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 1.8× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.69% | 99.79% |
| Error-free documents | 97.33% | 98.67% |
| Per 1,000 documents | $0.35 | $0.20 |
| To fix / 1,000 | 27 | 13 |
| Speed (median) | 2.1 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)
Both are equally accurate. GPT-6 Luna is 1.5× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 100.00% |
| Error-free documents | 100.00% | 100.00% |
| Per 1,000 documents | $0.40 | $0.27 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 2.4 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)
Both are equally accurate. GPT-6 Luna is 1.8× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 100.00% |
| Error-free documents | 100.00% | 100.00% |
| Per 1,000 documents | $0.48 | $0.27 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 2.7 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)
Both are equally accurate. GPT-6 Luna is 1.6× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.95% | 100.00% |
| Error-free documents | 99.33% | 100.00% |
| Per 1,000 documents | $0.42 | $0.26 |
| To fix / 1,000 | 7 | 0 |
| Speed (median) | 2.3 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)
Claude Haiku 5.5 is more accurate by 0.1 pp. GPT-6 Luna is 1.7× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.38% | 99.33% |
| Error-free documents | 92.00% | 91.33% |
| Per 1,000 documents | $0.45 | $0.27 |
| To fix / 1,000 | 80 | 87 |
| Speed (median) | 2.6 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 1.7 pp. GPT-6 Luna is 2.1× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 98.33% | 100.00% |
| Error-free documents | 90.00% | 100.00% |
| Per 1,000 documents | $0.34 | $0.16 |
| To fix / 1,000 | 100 | 0 |
| Speed (median) | 2.7 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 4.1 pp. GPT-6 Luna is 1.5× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 95.89% | 100.00% |
| Error-free documents | 77.33% | 100.00% |
| Per 1,000 documents | $0.32 | $0.21 |
| To fix / 1,000 | 227 | 0 |
| Speed (median) | 2.1 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 0.6 pp. GPT-6 Luna is 2.4× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.44% | 100.00% |
| Error-free documents | 96.67% | 100.00% |
| Per 1,000 documents | $0.46 | $0.19 |
| To fix / 1,000 | 33 | 0 |
| Speed (median) | 3.5 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 2.9 pp. GPT-6 Luna is 3.3× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 94.11% | 97.00% |
| Error-free documents | 64.67% | 82.00% |
| Per 1,000 documents | $0.61 | $0.19 |
| To fix / 1,000 | 353 | 180 |
| Speed (median) | 3.8 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.
Document summaries · French (150 documents)
GPT-6 Luna is more accurate by 6 pp. GPT-6 Luna is 3.4× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 92.78% | 98.78% |
| Error-free documents | 56.67% | 92.67% |
| Per 1,000 documents | $0.66 | $0.20 |
| To fix / 1,000 | 433 | 73 |
| Speed (median) | 4.0 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.
Questions about a document · English (US) (150 documents)
Claude Haiku 5.5 is more accurate by 0.9 pp. GPT-6 Luna is 1.5× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.11% |
| Error-free documents | 100.00% | 96.00% |
| Per 1,000 documents | $0.23 | $0.15 |
| To fix / 1,000 | 0 | 40 |
| Speed (median) | 1.8 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)
GPT-6 Luna is more accurate by 0.2 pp. GPT-6 Luna is 1.3× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.78% | 100.00% |
| Error-free documents | 98.67% | 100.00% |
| Per 1,000 documents | $0.30 | $0.23 |
| To fix / 1,000 | 13 | 0 |
| Speed (median) | 1.9 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)
Claude Haiku 5.5 is more accurate by 2 pp. GPT-6 Luna is 1.6× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 98.00% |
| Error-free documents | 100.00% | 91.33% |
| Per 1,000 documents | $0.32 | $0.20 |
| To fix / 1,000 | 0 | 87 |
| Speed (median) | 1.9 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 · Spanish (150 documents)
Claude Haiku 5.5 is more accurate by 0.3 pp. GPT-6 Luna is 1.5× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 100.00% | 99.67% |
| Error-free documents | 100.00% | 98.67% |
| Per 1,000 documents | $0.29 | $0.19 |
| To fix / 1,000 | 0 | 13 |
| Speed (median) | 1.8 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)
GPT-6 Luna is more accurate by 0.3 pp. GPT-6 Luna is 1.4× cheaper.
| Metric | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Correct fields | 99.22% | 99.56% |
| Error-free documents | 96.67% | 98.00% |
| Per 1,000 documents | $0.28 | $0.20 |
| To fix / 1,000 | 33 | 20 |
| Speed (median) | 1.8 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.