ModelVerdict

Comparison

Gemini 3.5 Flash Lite vs GPT-6 Sol

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; Gemini 3.5 Flash Lite is cheaper in 30 of 30 combinations.

More accurate

GPT-6 Sol

wins 23 of 30

Cheaper

Gemini 3.5 Flash Lite

cheaper in 30 of 30

Faster

Gemini 3.5 Flash Lite

0.9 sec vs 2.9 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

English (US)
Accuracy (correct fields)Per 1,000 documents
Task · documentsGemini 3.5 Flash LiteGPT-6 SolDifferenceGemini 3.5 Flash LiteGPT-6 SolDifference
Invoice extraction99.21%99.94%▲ 0.73$0.87$9.65+1014 %
Email triage96.30%99.04%▲ 2.74$0.42$2.44+482 %
Personal data detection99.52%99.71%▲ 0.19$0.41$2.09+413 %
Contract clauses99.95%100.00%▲ 0.05$0.69$3.33+381 %
Document summaries99.22%99.56%▲ 0.34$0.46$3.85+728 %
Questions about a document100.00%100.00%±0$0.46$2.36+411 %
Czech
Accuracy (correct fields)Per 1,000 documents
Task · documentsGemini 3.5 Flash LiteGPT-6 SolDifferenceGemini 3.5 Flash LiteGPT-6 SolDifference
Invoice extraction99.49%99.90%▲ 0.41$1.03$10.12+881 %
Email triage95.78%98.96%▲ 3.18$0.51$3.19+527 %
Personal data detection98.10%99.71%▲ 1.61$0.45$2.22+389 %
Contract clauses99.85%100.00%▲ 0.15$0.85$4.62+443 %
Document summaries99.56%100.00%▲ 0.44$0.69$4.72+581 %
Questions about a document100.00%100.00%±0$0.58$3.95+576 %
German
Accuracy (correct fields)Per 1,000 documents
Task · documentsGemini 3.5 Flash LiteGPT-6 SolDifferenceGemini 3.5 Flash LiteGPT-6 SolDifference
Invoice extraction100.00%99.95%▼ 0.05$1.03$10.65+934 %
Email triage94.67%98.22%▲ 3.55$0.48$2.73+472 %
Personal data detection99.62%99.71%▲ 0.09$0.45$2.13+368 %
Contract clauses98.00%100.00%▲ 2$0.79$4.36+451 %
Document summaries99.89%100.00%▲ 0.11$0.59$4.11+596 %
Questions about a document100.00%100.00%±0$0.52$3.23+525 %
Spanish
Accuracy (correct fields)Per 1,000 documents
Task · documentsGemini 3.5 Flash LiteGPT-6 SolDifferenceGemini 3.5 Flash LiteGPT-6 SolDifference
Invoice extraction98.21%99.23%▲ 1.02$1.05$10.79+928 %
Email triage95.56%98.44%▲ 2.88$0.48$2.77+477 %
Personal data detection95.81%99.81%▲ 4$0.45$2.21+396 %
Contract clauses99.79%100.00%▲ 0.21$0.84$4.48+436 %
Document summaries95.33%97.89%▲ 2.56$0.56$5.02+802 %
Questions about a document100.00%100.00%±0$0.51$3.36+556 %
French
Accuracy (correct fields)Per 1,000 documents
Task · documentsGemini 3.5 Flash LiteGPT-6 SolDifferenceGemini 3.5 Flash LiteGPT-6 SolDifference
Invoice extraction98.97%99.90%▲ 0.93$1.07$11.00+929 %
Email triage96.15%98.59%▲ 2.44$0.49$2.81+474 %
Personal data detection94.95%99.71%▲ 4.76$0.46$2.23+389 %
Contract clauses99.95%99.59%▼ 0.36$0.85$4.49+426 %
Document summaries94.67%97.22%▲ 2.55$0.65$4.58+599 %
Questions about a document99.56%100.00%▲ 0.44$0.53$3.43+542 %

Difference: GPT-6 Sol 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 Sol is more accurate by 0.7 pp. Gemini 3.5 Flash Lite is 11.1× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.21%99.94%
Error-free documents92.00%99.33%
Per 1,000 documents$0.87$9.65
To fix / 1,000807
Speed (median)1.7 sec4.1 sec
Valid answers100.00%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)

GPT-6 Sol is more accurate by 0.4 pp. Gemini 3.5 Flash Lite is 9.8× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.49%99.90%
Error-free documents93.33%98.67%
Per 1,000 documents$1.03$10.12
To fix / 1,0006713
Speed (median)1.8 sec3.9 sec
Valid answers100.00%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)

Both are equally accurate. Gemini 3.5 Flash Lite is 10.3× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields100.00%99.95%
Error-free documents100.00%99.33%
Per 1,000 documents$1.03$10.65
To fix / 1,00007
Speed (median)1.5 sec4.9 sec
Valid answers100.00%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 Sol is more accurate by 1 pp. Gemini 3.5 Flash Lite is 10.3× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields98.21%99.23%
Error-free documents78.00%90.00%
Per 1,000 documents$1.05$10.79
To fix / 1,000220100
Speed (median)1.6 sec5.9 sec
Valid answers100.00%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 Sol is more accurate by 0.9 pp. Gemini 3.5 Flash Lite is 10.3× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields98.97%99.90%
Error-free documents88.00%98.67%
Per 1,000 documents$1.07$11.00
To fix / 1,00012013
Speed (median)1.5 sec6.0 sec
Valid answers100.00%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 Sol is more accurate by 2.7 pp. Gemini 3.5 Flash Lite is 5.8× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields96.30%99.04%
Error-free documents68.67%92.00%
Per 1,000 documents$0.42$2.44
To fix / 1,00031380
Speed (median)0.7 sec2.8 sec
Valid answers100.00%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 3.2 pp. Gemini 3.5 Flash Lite is 6.3× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields95.78%98.96%
Error-free documents64.00%90.67%
Per 1,000 documents$0.51$3.19
To fix / 1,00036093
Speed (median)0.7 sec2.8 sec
Valid answers100.00%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 3.5 pp. Gemini 3.5 Flash Lite is 5.7× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields94.67%98.22%
Error-free documents56.67%84.00%
Per 1,000 documents$0.48$2.73
To fix / 1,000433160
Speed (median)0.8 sec2.9 sec
Valid answers100.00%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 2.9 pp. Gemini 3.5 Flash Lite is 5.8× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields95.56%98.44%
Error-free documents65.33%86.00%
Per 1,000 documents$0.48$2.77
To fix / 1,000347140
Speed (median)0.8 sec3.3 sec
Valid answers100.00%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 2.4 pp. Gemini 3.5 Flash Lite is 5.7× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields96.15%98.59%
Error-free documents68.67%87.33%
Per 1,000 documents$0.49$2.81
To fix / 1,000313127
Speed (median)0.8 sec3.2 sec
Valid answers100.00%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 Sol is more accurate by 0.2 pp. Gemini 3.5 Flash Lite is 5.1× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.52%99.71%
Error-free documents97.33%98.00%
Per 1,000 documents$0.41$2.09
To fix / 1,0002720
Speed (median)0.8 sec2.6 sec
Valid answers100.00%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 1.6 pp. Gemini 3.5 Flash Lite is 4.9× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields98.10%99.71%
Error-free documents88.00%98.00%
Per 1,000 documents$0.45$2.22
To fix / 1,00012020
Speed (median)0.8 sec2.5 sec
Valid answers100.00%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)

GPT-6 Sol is more accurate by 0.1 pp. Gemini 3.5 Flash Lite is 4.7× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.62%99.71%
Error-free documents97.33%98.00%
Per 1,000 documents$0.45$2.13
To fix / 1,0002720
Speed (median)0.8 sec2.6 sec
Valid answers100.00%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 Sol is more accurate by 4 pp. Gemini 3.5 Flash Lite is 5× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields95.81%99.81%
Error-free documents73.33%98.67%
Per 1,000 documents$0.45$2.21
To fix / 1,00026713
Speed (median)0.9 sec2.9 sec
Valid answers100.00%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 Sol is more accurate by 4.8 pp. Gemini 3.5 Flash Lite is 4.9× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields94.95%99.71%
Error-free documents67.33%98.00%
Per 1,000 documents$0.46$2.23
To fix / 1,00032720
Speed (median)0.8 sec2.9 sec
Valid answers100.00%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)

Both are equally accurate. Gemini 3.5 Flash Lite is 4.8× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.95%100.00%
Error-free documents99.33%100.00%
Per 1,000 documents$0.69$3.33
To fix / 1,00070
Speed (median)1.0 sec2.5 sec
Valid answers100.00%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. Gemini 3.5 Flash Lite is 5.4× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.85%100.00%
Error-free documents98.00%100.00%
Per 1,000 documents$0.85$4.62
To fix / 1,000200
Speed (median)0.9 sec2.0 sec
Valid answers100.00%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. Gemini 3.5 Flash Lite is 5.5× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields98.00%100.00%
Error-free documents84.00%100.00%
Per 1,000 documents$0.79$4.36
To fix / 1,0001600
Speed (median)1.0 sec2.9 sec
Valid answers100.00%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 Sol is more accurate by 0.2 pp. Gemini 3.5 Flash Lite is 5.4× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.79%100.00%
Error-free documents98.00%100.00%
Per 1,000 documents$0.84$4.48
To fix / 1,000200
Speed (median)1.1 sec2.8 sec
Valid answers100.00%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 0.4 pp. Gemini 3.5 Flash Lite is 5.3× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.95%99.59%
Error-free documents99.33%94.67%
Per 1,000 documents$0.85$4.49
To fix / 1,000753
Speed (median)1.0 sec2.6 sec
Valid answers100.00%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 Sol is more accurate by 0.3 pp. Gemini 3.5 Flash Lite is 8.3× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.22%99.56%
Error-free documents95.33%97.33%
Per 1,000 documents$0.46$3.85
To fix / 1,0004727
Speed (median)0.9 sec8.1 sec
Valid answers100.00%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 0.4 pp. Gemini 3.5 Flash Lite is 6.8× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.56%100.00%
Error-free documents97.33%100.00%
Per 1,000 documents$0.69$4.72
To fix / 1,000270
Speed (median)1.0 sec7.6 sec
Valid answers100.00%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 Sol is more accurate by 0.1 pp. Gemini 3.5 Flash Lite is 7× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.89%100.00%
Error-free documents99.33%100.00%
Per 1,000 documents$0.59$4.11
To fix / 1,00070
Speed (median)0.9 sec7.1 sec
Valid answers100.00%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 Sol is more accurate by 2.6 pp. Gemini 3.5 Flash Lite is 9× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields95.33%97.89%
Error-free documents72.00%87.33%
Per 1,000 documents$0.56$5.02
To fix / 1,000280127
Speed (median)0.9 sec7.0 sec
Valid answers100.00%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 Sol is more accurate by 2.5 pp. Gemini 3.5 Flash Lite is 7× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields94.67%97.22%
Error-free documents71.33%83.33%
Per 1,000 documents$0.65$4.58
To fix / 1,000287167
Speed (median)0.9 sec6.1 sec
Valid answers100.00%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)

Both are equally accurate. Gemini 3.5 Flash Lite is 5.1× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.46$2.36
To fix / 1,00000
Speed (median)0.8 sec2.3 sec
Valid answers100.00%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. Gemini 3.5 Flash Lite is 6.8× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.58$3.95
To fix / 1,00000
Speed (median)0.8 sec2.3 sec
Valid answers100.00%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)

Both are equally accurate. Gemini 3.5 Flash Lite is 6.2× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.52$3.23
To fix / 1,00000
Speed (median)0.7 sec2.2 sec
Valid answers100.00%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. Gemini 3.5 Flash Lite is 6.6× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.51$3.36
To fix / 1,00000
Speed (median)0.8 sec2.5 sec
Valid answers100.00%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 0.4 pp. Gemini 3.5 Flash Lite is 6.4× cheaper.

MetricGemini 3.5 Flash LiteGPT-6 Sol
Correct fields99.56%100.00%
Error-free documents98.00%100.00%
Per 1,000 documents$0.53$3.43
To fix / 1,000200
Speed (median)0.8 sec2.5 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Log in with email

We use your email only to log you in (and for the digest if you tick it). No spam.