All the models in one place
Run Ideogram V4.5 and every other model on Glif from one chat. One account, one balance, no API keys to wire up.
Ideogram V4.5
Quality and cost
Product photography leaderboardProduct photography, against 14 other ranked models
- This model (Ideogram V4.5)
- Current default
Predicted human preference within this mode, not an absolute quality grade. Cost is the midpoint of each estimated range, not a quote. Hover or focus for raw ratings.
Show score details
- 1GPT Image 2.5Rating 1992 · 90% range 1872–2145 · 23 games94.4% · 0.7-73 credits
- 2Gemini 3.1 Flash Image GeneratorRating 1752 · 90% range 1631–1888 · 29 games81.0% · 6-20 credits
- 3FLUX.2 Max Image GeneratorRating 1747 · 90% range 1642–1899 · 21 games80.6% · up to 115 credits
- 4FLUX.2 Flex Image GeneratorRating 1626 · 90% range 1544–1732 · 24 games67.4% · up to 99 credits
- 5Seedream 5.0 ProRating 1588 · 90% range 1493–1713 · 22 games62.4% · 6-16 credits
- 6Phota Image EditorRating 1515 · 90% range 1416–1607 · 21 games52.2% · 12-24 credits
- 7Nano Banana Pro Image GeneratorRating 1503 · 90% range 1409–1615 · 22 games50.4% · 18-34 credits
- 8Ideogram V4.5Rating 1472 · 90% range 1361–1572 · 32 games46.0% · 1-30 credits
- 9Seedream V4.5Rating 1441 · 90% range 1342–1560 · 22 games41.6% · 5 credits
- 10Muse ImageRating 1399 · 90% range 1296–1481 · 23 games35.9% · 1 credit
- 11Grok Imagine ImageRating 1324 · 90% range 1238–1398 · 23 games26.6% · 3 credits
- 12MAI Image 2.5 GeneratorRating 1318 · 90% range 1233–1403 · 21 games26.0% · 7-8 credits
- 13Z-Image GeneratorRating 1306 · 90% range 1186–1393 · 22 games24.7% · up to 7 credits
- 14FLUX.2 Pro Image GeneratorRating 1287 · 90% range 1211–1349 · 22 games22.7% · up to 49 credits
- 15Seedream 5.0 LiteRating 1230 · 90% range 1131–1316 · 23 games17.4% · 5 credits
Head-to-head ratings
Blind votes by the Glif team, as of
- Product photography Elo
- 1472
- 90% range 1361–1572
- Rank
- #8 of 15
- Games
- 32
Elo rating centred on 1500, compared only within this mode.
Show score details
46.0% predicted preference against a 1500-rated reference · Wins 10 / Losses 13 / Ties 9.
Bradley–Terry ratings center on 1500 and have no maximum. Preference = 100 / (1 + 10^((1500 − rating) / 400)). LLM-judge scores are not included in this export.
Other modes
Character consistency1441 Elo · #9 of 14
Text-to-image1670 Elo · #2 of 13
Brand and text renderingNeeds more ratings (16 of 20 games)
Test outputs
Compare all modelsMore from the Ideogram family
Explore other options from the same family.