Updated March 26, 2026· Based on independent benchmark data
GPT-5.4 (xhigh) leads in intelligence with a score of 57.2 vs 11.9. Llama 3.2 Instruct 90B (Vision) is 3.5x cheaper at $0.72/1M tokens vs $2.50/1M. For speed, GPT-5.4 (xhigh) wins at 77 tok/s vs 42 tok/s.
| Metric | Llama 3.2 Instruct 90B (Vision) | GPT-5.4 (xhigh) |
|---|---|---|
| Intelligence Score | 11.9 | 57.2 |
| Coding Score | N/A | 57.3 |
| Math Score | N/A | N/A |
| Speed (tok/s) | 42 tok/s | 77 tok/s |
| Latency (TTFT) | 0.38s | 181.82s |
| Input Price / 1M tokens | $0.72 | $2.50 |
| Output Price / 1M tokens | $0.72 | $15 |
| Context Window | N/A | N/A |
GPT-5.4 (xhigh) outperforms Llama 3.2 Instruct 90B (Vision) on the intelligence index with a score of 57.2 compared to 11.9.
GPT-5.4 (xhigh) generates output significantly faster at 77 tok/s compared to Llama 3.2 Instruct 90B (Vision)'s 42 tok/s, making it 1.8x faster for streaming responses. Time to first token is 0.38s for Llama 3.2 Instruct 90B (Vision) vs 181.82s for GPT-5.4 (xhigh), which affects perceived responsiveness in interactive applications.
Llama 3.2 Instruct 90B (Vision) is more affordable at $0.72/1M input tokens ($0.72/1M output), while GPT-5.4 (xhigh) costs $2.50/1M input ($15/1M output). That makes GPT-5.4 (xhigh) 3.5x more expensive per token, which can add up significantly at scale. For a typical workload of 100 requests per day at 2,000 tokens each, Llama 3.2 Instruct 90B (Vision) would cost approximately $4.32/month vs $15.00/month for GPT-5.4 (xhigh) in input costs alone.
Choose Llama 3.2 Instruct 90B (Vision) when you need lower cost. Choose GPT-5.4 (xhigh) when you need higher intelligence (57.2), faster output (77 tok/s).
Llama 3.2 Instruct 90B (Vision) is cheaper at $0.72/1M input tokens vs $2.50/1M for GPT-5.4 (xhigh).
GPT-5.4 (xhigh) is faster, producing output at 77 tok/s compared to Llama 3.2 Instruct 90B (Vision)'s 42 tok/s.
No, Llama 3.2 Instruct 90B (Vision) does not support image input. Neither model supports image input.
It depends on your priorities. GPT-5.4 (xhigh) scores higher on intelligence (57.2), but Llama 3.2 Instruct 90B (Vision) may be better for specific use cases like budget-conscious projects or speed-critical applications.
Data last synced: March 26, 2026
| Max Output Tokens | N/A | N/A |
| Input Modalities | Text | Text |
| Output Modalities | Text | Text |
| Free Tier | No | No |