Best AI model for ASCII art
In ASCII art, work within 60 Elo of the best model is available for 6× less: Muse Spark 1.2 at $1.98 per thousand tasks against $11.25 for Claude Opus 5.
63 models rated head to head in this category, not scored by a general benchmark and assumed to transfer. Figures and diagrams composed purely from characters, a strict test of spatial reasoning without any rendering to fall back on.
Nothing cheaper is as good
4 models are undominated here: for each, no cheaper model scores within 18 Elo. Everything else in the table below is beaten on both counts at once.
- 1DeepSeek V4 Flash 04231112 Elo$0.170 / 1k1.05M contextfp4 on some hosts
- 2MiMo-V2.51143 Elo$0.183 / 1k1.05M context
- 3GLM 5.3 Flash1266 Elo$0.230 / 1k1.05M contextfp4 on some hosts
- 4Claude Opus 51346 Elo$11.25 / 1k1.00M context
Every model rated on ASCII art
Elo measured in this category specifically. A model strong overall can sit well down this table, which is the whole reason for ranking by task.
| # | Model | Elo | $/1k | Context | Hosts |
|---|---|---|---|---|---|
| 1 | Claude Opus 5 | 1346 | $11.25 | 1.00M | 11 |
| 2 | Claude Fable 5 | 1329 | $20.52 | 1.00M | 6 |
| 3 | Claude Fable 5.1 | 1324 | $19.98 | 1.00M | 4 |
| 4 | Muse Spark 1.2 | 1293 | $1.98 | 1.05M | 1 |
| 5 | Grok 4.6 | 1280 | $3.40 | 500K | 6 |
| 6 | Muse Spark 1.1 | 1279 | $1.98 | 1.05M | 1 |
| 7 | Gemini 3.6 Flash | 1277 | $1.54 | 1.05M | 7 |
| 8 | Claude Opus 4.8 | 1277 | $11.25 | 1.00M | 11 |
| 9 | Gemini 3.7 Flash | 1274 | $1.54 | 1.05M | 6 |
| 10 | Gemini 3.1 Pro Preview | 1271 | $4.70 | 1.05M | 6 |
| 11 | GLM 5.3 Flash | 1266 | $0.230 | 1.05M | 32 |
| 12 | Gemini 3.8 Flash | 1264 | $1.54 | 1.05M | 6 |
| 13 | GPT-5.5 | 1256 | $12.90 | 1.05M | 7 |
| 14 | Claude Opus 4.6 | 1256 | $10.26 | 1.00M | 6 |
| 15 | Grok 4.5 | 1254 | $4.36 | 500K | 4 |
| 16 | Gemini 3.5 Flash | 1254 | $3.53 | 1.05M | 7 |
| 17 | Muse Spark 1.3 | 1253 | $1.98 | 1.05M | 1 |
| 18 | Claude Sonnet 4.6 | 1236 | $6.16 | 1.00M | 9 |
| 19 | Qwen3.7 Max | 1214 | $2.61 | 1.00M | 2 |
| 20 | Claude Sonnet 5 | 1208 | $4.30 | 1.00M | 10 |
| 21 | Claude Sonnet 4.5 | 1206 | $6.16 | 1.00M | 7 |
| 22 | GLM 5.3 | 1203 | $1.93 | 1.05M | 41 |
| 23 | Kimi K2.7 Code | 1200 | $1.75 | 262K | 12 |
| 24 | GPT-5.4 | 1200 | $6.45 | 1.05M | 7 |
| 25 | Mistral Large 4 | 1197 | $1.00 | 1.05M | 3 |
| 26 | GLM 5.2 | 1194 | $2.00 | 1.05M | 34 |
| 27 | Claude Opus 4.5 | 1191 | $10.26 | 200K | 6 |
| 28 | Grok 4.20 | 1176 | $2.17 | 2.00M | 4 |
| 29 | Claude Opus 4.1 | 1168 | $30.78 | 200K | 1 |
| 30 | Gemini 3.1 Flash Lite Preview | 1167 | $0.588 | 1.05M | 3 |
| 31 | Kimi K2.5 | 1167 | $1.22 | 262K | 5 |
| 32 | Kimi K2.6 | 1166 | $1.52 | 262K | 17 |
| 33 | GLM 4.7 | 1166 | $0.980 | 203K | 6 |
| 34 | MiniMax M3 | 1163 | $0.547 | 1.05M | 13 |
| 35 | GLM 5 Turbo | 1159 | $1.95 | 203K | 1 |
| 36 | MiMo-V2.5-Pro | 1155 | $0.497 | 1.05M | 6 |
| 37 | Step 3.7 Flash | 1153 | $0.470 | 262K | 2 |
| 38 | GPT-5.2 | 1153 | $5.17 | 400K | 4 |
| 39 | GPT-5.3-Codex | 1152 | $5.17 | 400K | 3 |
| 40 | DeepSeek V4 Pro 0423 | 1150 | $1.68 | 1.05M | 15 |
| 41 | GLM 5 | 1148 | $1.42 | 203K | 8 |
| 42 | GLM 5.1 | 1145 | $2.11 | 203K | 13 |
| 43 | MiMo-V2.5 | 1143 | $0.183 | 1.05M | 5 |
| 44 | GPT-5 | 1142 | $3.69 | 400K | 3 |
| 45 | Qwen3.7 Plus | 1141 | $0.584 | 1.00M | 2 |
| 46 | Claude Haiku 4.5 | 1141 | $2.15 | 200K | 8 |
| 47 | MiniMax M2.7 | 1138 | $0.534 | 197K | 6 |
| 48 | Grok 4.3 | 1130 | $2.17 | 1.00M | 4 |
| 49 | GPT-5 Mini | 1124 | $0.738 | 400K | 4 |
| 50 | GLM 5V Turbo | 1119 | $1.95 | 203K | 1 |
| 51 | GPT-5.1 | 1119 | $3.69 | 400K | 5 |
| 52 | Qwen3.6 Plus | 1115 | $0.975 | 1.00M | 1 |
| 53 | DeepSeek V4 Flash 0423 | 1112 | $0.170 | 1.05M | 16 |
| 54 | GPT-5.1-Codex-Mini | 1106 | $0.742 | 400K | 1 |
| 55 | Nemotron 3 Ultra | 1098 | $1.09 | 203K | 4 |
| 56 | Qwen3.5 Plus 2026-02-15 | 1095 | $0.780 | 1.00M | 1 |
| 57 | Inkling | 1092 | $1.81 | 524K | 2 |
| 58 | DeepSeek V4 Flash 0731 | 1091 | $0.233 | 1.05M | 27 |
| 59 | DeepSeek V3.2 | 1084 | $0.504 | 164K | 12 |
| 60 | Mistral Large 3 2512 | 1070 | $0.760 | 262K | 2 |
| 61 | Trinity Large Thinking | 1042 | $0.403 | 262K | 1 |
| 62 | Mercury 2 | 1000 | $0.363 | 128K | 1 |
| 63 | Mistral Medium 3.1 | 998 | $0.821 | 131K | 3 |
Other tasks
Elo comes from head-to-head build-offs via OpenRouter, compiled 12 October 2026 and refreshed daily. Cost is the median across every host serving each model, priced on 1,200 input and 300 output tokens with 60% cached. See the findings for method, or download the data.