Updated March 26, 2026· Based on independent benchmark data
GPT-5 nano (high) leads in intelligence with a score of 26.8 vs 24.5. For speed, gpt-oss-20B (high) wins at 281 tok/s vs 132 tok/s.
| Metric | GPT-5 nano (high) | gpt-oss-20B (high) |
|---|---|---|
| Intelligence Score | 26.8 | 24.5 |
| Coding Score | 20.3 | 18.5 |
| Math Score | 83.7 | 89.3 |
| Speed (tok/s) | 132 tok/s | 281 tok/s |
| Latency (TTFT) | 99.68s | 0.48s |
| Input Price / 1M tokens | $0.05 | $0.06 |
| Output Price / 1M tokens | $0.40 | $0.20 |
| Context Window | N/A |
GPT-5 nano (high) outperforms gpt-oss-20B (high) on the intelligence index with a score of 26.8 compared to 24.5. Their coding capabilities are closely matched (20.3 vs 18.5). In mathematical reasoning, gpt-oss-20B (high) leads with 89.3 compared to GPT-5 nano (high)'s 83.7.
gpt-oss-20B (high) generates output significantly faster at 281 tok/s compared to GPT-5 nano (high)'s 132 tok/s, making it 2.1x faster for streaming responses. Time to first token is 0.48s for gpt-oss-20B (high) vs 99.68s for GPT-5 nano (high), which affects perceived responsiveness in interactive applications.
GPT-5 nano (high) is more affordable at $0.05/1M input tokens ($0.40/1M output), while gpt-oss-20B (high) costs $0.06/1M input ($0.20/1M output). For a typical workload of 100 requests per day at 2,000 tokens each, GPT-5 nano (high) would cost approximately $0.30/month vs $0.36/month for gpt-oss-20B (high) in input costs alone.
Choose GPT-5 nano (high) when you need higher intelligence (26.8). Choose gpt-oss-20B (high) when you need faster output (281 tok/s).
GPT-5 nano (high) scores higher on coding benchmarks (20.3 vs 18.5), making it the better choice for programming tasks.
GPT-5 nano (high) is cheaper at $0.05/1M input tokens vs $0.06/1M for gpt-oss-20B (high).
gpt-oss-20B (high) is faster, producing output at 281 tok/s compared to GPT-5 nano (high)'s 132 tok/s.
No, GPT-5 nano (high) does not support image input. Neither model supports image input.
Data last synced: March 26, 2026
| N/A |
| Max Output Tokens | N/A | N/A |
| Input Modalities | Text | Text |
| Output Modalities | Text | Text |
| Free Tier | No | No |
It depends on your priorities. GPT-5 nano (high) scores higher on intelligence (26.8), but gpt-oss-20B (high) may be better for specific use cases like budget-conscious projects or speed-critical applications.