Qwen3.5 122B A10B is cheaper than Llama 3.1 70B (Instruct) at $0.4/M vs $0.59/M input tokens.
| Spec | Llama 3.1 70B (Instruct) | Qwen3.5 122B A10B |
|---|---|---|
| Input price | $0.59/M tokens | $0.4/M tokens |
| Output price | $0.79/M tokens | $3.2/M tokens |
| Context window | 128K tokens | 262K tokens |
| Parameters | 70B | 122B (10B active) |
| Weights | Open | Open |
| Free API (OpenRouter) | No | No |
| Released | Jul 2024 | Feb 2026 |
Not enough votes to call it. On the specs, nothing separates them.
Llama 3.1 70B (Instruct) costs 4.1x less per token.
Qwen3.5 122B A10B wins input (1.5× cheaper)·Llama 3.1 70B (Instruct) wins output (4.1× cheaper)
Per million tokens. Prices and uptime via OpenRouter, checked 9 Sep 2026.
Qwen3.5 122B A10B uses 10.4x more emoji