Side-by-side comparison of pricing, 12 benchmarks, and generation speed.
| Metric | Qwen3.5 0.8B (Non-reasoning) | Llama 3.2 Instruct 3B |
|---|---|---|
| Input ($/M tokens) | $0.01 | $0.085 |
| Output ($/M tokens) | $0.05 | $0.085 |
Data from Artificial Analysis API — 12 benchmarks
Qwen3.5 0.8B (Non-reasoning) is cheaper overall. Its blended price (3:1 input/output ratio) is $0.02/M tokens vs $0.09/M for Llama 3.2 Instruct 3B.
Llama 3.2 Instruct 3B wins 9 out of 12 benchmarks compared to 3 for Qwen3.5 0.8B (Non-reasoning). See the detailed benchmark chart above for per-category results.
Qwen3.5 0.8B (Non-reasoning) generates tokens faster at 306 tok/s vs 54 tok/s. Qwen3.5 0.8B (Non-reasoning) also has lower time-to-first-token (0.23s vs 0.35s).
Choose based on your priorities: Qwen3.5 0.8B (Non-reasoning) for lower cost, Llama 3.2 Instruct 3B for stronger benchmark performance, and Qwen3.5 0.8B (Non-reasoning) for faster generation. For latency-sensitive apps, check the TTFT comparison above.