Updated March 26, 2026· Based on independent benchmark data
Qwen3.5 397B A17B (Reasoning) and GPT-5 Codex (high) are virtually tied on intelligence (45.0 vs 44.6). Qwen3.5 397B A17B (Reasoning) is 2.1x cheaper at $0.60/1M tokens vs $1.25/1M. For speed, GPT-5 Codex (high) wins at 180 tok/s vs 53 tok/s.
| Metric | Qwen3.5 397B A17B (Reasoning) | GPT-5 Codex (high) |
|---|---|---|
| Intelligence Score | 45.0 | 44.6 |
| Coding Score | 41.3 | 38.9 |
| Math Score | N/A | 98.7 |
| Speed (tok/s) | 53 tok/s | 180 tok/s |
| Latency (TTFT) | 1.46s | 9.17s |
| Input Price / 1M tokens | $0.60 | $1.25 |
| Output Price / 1M tokens | $3.60 | $10 |
| Context Window | N/A |
Qwen3.5 397B A17B (Reasoning) and GPT-5 Codex (high) perform similarly on overall intelligence, scoring 45.0 and 44.6 respectively. For coding tasks, Qwen3.5 397B A17B (Reasoning) has the edge with a coding score of 41.3 vs 38.9.
GPT-5 Codex (high) generates output significantly faster at 180 tok/s compared to Qwen3.5 397B A17B (Reasoning)'s 53 tok/s, making it 3.4x faster for streaming responses. Time to first token is 1.46s for Qwen3.5 397B A17B (Reasoning) vs 9.17s for GPT-5 Codex (high), which affects perceived responsiveness in interactive applications.
Qwen3.5 397B A17B (Reasoning) is more affordable at $0.60/1M input tokens ($3.60/1M output), while GPT-5 Codex (high) costs $1.25/1M input ($10/1M output). That makes GPT-5 Codex (high) 2.1x 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, Qwen3.5 397B A17B (Reasoning) would cost approximately $3.60/month vs $7.50/month for GPT-5 Codex (high) in input costs alone.
Choose Qwen3.5 397B A17B (Reasoning) when you need stronger coding performance (41.3), lower cost. Choose GPT-5 Codex (high) when you need faster output (180 tok/s).
Qwen3.5 397B A17B (Reasoning) scores higher on coding benchmarks (41.3 vs 38.9), making it the better choice for programming tasks.
Qwen3.5 397B A17B (Reasoning) is cheaper at $0.60/1M input tokens vs $1.25/1M for GPT-5 Codex (high).
GPT-5 Codex (high) is faster, producing output at 180 tok/s compared to Qwen3.5 397B A17B (Reasoning)'s 53 tok/s.
No, Qwen3.5 397B A17B (Reasoning) 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 |
Both models perform similarly on intelligence benchmarks. Choose based on specific needs: pricing, speed, context window, or provider ecosystem.