gemini-3-flash-preview
Google- average±1.6 (95%)
- 82.6
- leaderboard rankof 114
- ≈14
- benchmarksof 12
- 12
- per 1,000 documentsmean over its benchmarks
- $44
- seconds per inputmean response time
- 66.4
By task
-
Data correction
59.6
Best of 114 models with all its benchmarks: claude-opus-5-5, 82.7.
-
Document understanding
91.2
Best of 107 models with all its benchmarks: gpt-6-astra, 97.0.
-
Information extraction
86.8
Best of 107 models with all its benchmarks: grok-4.6, 90.8.
-
Transcription
90.4
Best of 114 models with all its benchmarks: gemini-4-argon, 96.2.
By benchmark
| Benchmark | ScoreThe best current run on each benchmark, on the benchmark's metric scaled to 0–100. ± is the 95% sampling margin, from resampling the run's inputs. | RankRank among all current models on that benchmark; ≈ for models that cannot be told apart within sampling error. | Per 1,000 docs | s / input | Best run |
|---|---|---|---|---|---|
| Bibliographic Data | 89.1±3.0 | ≈74of 115 | $60$0.30 run | 108.76 | T0499 · 25 Jan 2026prompt.txt |
| Blacklist Cards | 92.2±3.3 | ≈51of 114 | $7.75$0.26 run | 11.98 | T0506 · 16 Sep 2026prompt.txt |
| Book Advert XML Repair | 23.7±7.6 | ≈64of 119 | $75$3.75 run | 88.26 | T0509 · 16 Sep 2026prompt.txt |
| Business Letters | 67.0±7.3 | ≈10of 114 | $20$1.12 run | 38.27 | T0500 · 25 Jan 2026prompt.txt |
| Company Lists | 89.5±8.5 | ≈6of 114 | $14$0.21 run | 19.31 | T0507 · 16 Sep 2026prompt.txt |
| Duty Rosters | 89.3±3.1 | ≈10of 107 | $56$1.17 run | 69.0 | T1284 · 16 Sep 2026prompt.txt |
| Fraktur Adverts | 97.0±4.0 | ≈15of 114 | $101$0.51 run | 197.42 | T0503 · 25 Jan 2026prompt_optimized.txt |
| General Meeting Minutes | 94.4±3.4 | ≈1of 108 | $57$0.52 run | 67.99 | T1231 · 16 Sep 2026prompt.txt |
| Library Cards | 85.9±1.5 | ≈51of 114 | $24$6.35 run | 43.36 | T0504 · 25 Jan 2026prompt.txt |
| Magazine Pages | 88.8±6.9 | ≈18of 115 | $6.77$0.31 run | 11.52 | T0804 · 16 Sep 2026prompt.txt |
| Medieval Manuscripts | 78.7±9.2 | ≈27of 114 | $83$0.99 run | 95.19 | T0505 · 16 Sep 2026prompt.txt |
| Personnel Cards | 95.5±1.7 | ≈58of 114 | $26$1.59 run | 45.77 | T0593 · 9 Feb 2026prompt.txt |
Price
| From | Input, $ per million tokens | Output, $ per million tokens | Source |
|---|---|---|---|
| 23 Jan 2026 | 0.5 | 3.0 | archived price page |
Every run
- T1284Duty Rosters16 Sep 202689.3$1.17
- T1231General Meeting Minutes16 Sep 202694.4$0.52
- T0804Magazine Pages16 Sep 202688.8$0.31
- T0593Personnel Cards16 Sep 202689.9$1.89
- T0509Book Advert XML Repair16 Sep 202623.7$3.75
- T0508Company Lists16 Sep 202662.2$0.18
- T0507Company Lists16 Sep 202689.5$0.21
- T0506Blacklist Cards16 Sep 202692.2$0.26
- T0505Medieval Manuscripts16 Sep 202678.7$0.99
- T0503Fraktur Adverts16 Sep 202639.6$0.37
- T0502Business Letters16 Sep 2026 · 32 signed letters68.0$0.41
- T0501Business Letters16 Sep 2026 · 25 unsigned letters78.0$0.15
- T0500Business Letters16 Sep 2026 · all 57 letters67.0$0.74
- T0499Bibliographic Data16 Sep 202674.6$0.23
- T1284Duty Rosters24 Jul 202688.9$1.74
- T1231General Meeting Minutes3 Jul 202692.7$0.028
- T0804Magazine Pages23 Mar 202684.8$0.86
- T0504Library Cards12 Mar 202685.4$9.62
- T0593Personnel Cards9 Feb 202695.5$1.59
- T0509Book Advert XML Repair25 Jan 202623.4$6.47
- T0508Company Lists25 Jan 202661.9$0.21
- T0507Company Lists25 Jan 202687.1$0.24
- T0506Blacklist Cards25 Jan 202690.8$0.22
- T0505Medieval Manuscripts25 Jan 202677.8$0.88
- T0504Library Cards25 Jan 202685.9$6.35
- T0503Fraktur Adverts25 Jan 202697.0$0.51
- T0502Business Letters25 Jan 2026 · 32 signed letters66.0$0.95
- T0501Business Letters25 Jan 2026 · 25 unsigned letters78.0$0.17
- T0500Business Letters25 Jan 2026 · all 57 letters67.0$1.12
- T0499Bibliographic Data25 Jan 202689.1$0.30