Roo Code tests each frontier model against a suite of hundreds of exercises across 5 programming languages with varying difficulty. These results can help you find the right price-to-intelligence ratio for your use case.
Want to see the results for a model we haven't tested yet? Ping us in Discord.
| Model | Metrics | Scores | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Name Context Window | Price In / Out | Duration | Tokens In / Out | Cost USD | Total | |||||
Claude Sonnet 4 0 | $0.00 / $0.00 | 5h 35m 31s | 39M / 644K | $39.61 | 94% | 100% | 98% | 100% | 97% | 98% |
Gemini 2.5 Pro 0 | $0.00 / $0.00 | 6h 17m 23s | 43M / 1M | $57.80 | 97% | 91% | 96% | 100% | 97% | 96% |
Claude Opus 4 0 | $0.00 / $0.00 | 7h 50m 29s | 30M / 485K | $172.29 | 92% | 91% | 94% | 94% | 100% | 94% |
Claude 3.7 Sonnet 0 | $0.00 / $0.00 | 4h 52m 36s | 19M / 603K | $27.16 | 92% | 93% | 98% | 97% | 87% | 94% |
GPT 4.1 0 | $0.00 / $0.00 | 4h 39m 51s | 37M / 624K | $38.64 | 92% | 91% | 90% | 94% | 90% | 91% |
Gemini 2.5 Flash 0 | $0.00 / $0.00 | 3h 39m 38s | 61M / 1M | $14.15 | 89% | 91% | 92% | 85% | 90% | 90% |
Claude 3.5 Sonnet 0 | $0.00 / $0.00 | 3h 37m 58s | 19M / 323K | $24.98 | 94% | 91% | 92% | 88% | 80% | 90% |
Grok 3 0 | $0.00 / $0.00 | 5h 14m 20s | 40M / 890K | $74.40 | 97% | 89% | 90% | 91% | 77% | 89% |
Qwen 3 Coder 0 | $0.00 / $0.00 | 7h 56m 14s | 51M / 828K | $27.63 | 86% | 80% | 82% | 85% | 87% | 84% |
Kimi K2 0 | $0.00 / $0.00 | 7h 52m 24s | 27M / 433K | $12.39 | 81% | 80% | 88% | 82% | 83% | 83% |
GPT 4.1 Mini 0 | $0.00 / $0.00 | 5h 17m 57s | 47M / 715K | $8.81 | 81% | 84% | 94% | 76% | 70% | 83% |
Qwen3 235B A22B 2507 0 | $0.00 / $0.00 | 8h 3m 37s | 44M / 498K | $6.94 | 69% | 84% | 82% | 79% | 80% | 79% |
o4 Mini (High) 0 | $0.00 / $0.00 | 14h 44m 26s | 13M / 3M | $25.70 | 75% | 82% | 86% | 79% | 67% | 79% |
DeepSeek V3 0 | $0.00 / $0.00 | 7h 12m 41s | 30M / 524K | $12.82 | 83% | 76% | 82% | 76% | 67% | 77% |
o3 Mini (High) 0 | $0.00 / $0.00 | 13h 1m 13s | 12M / 2M | $20.36 | 67% | 78% | 72% | 88% | 73% | 75% |