Head to head
Compare models
Pick 2–3 models to overlay their suite radars and line up every stat. Best value in each row is highlighted.
✓ Verified data — every figure is cited. See sources.
DeepSeek V3.2 (Thinking)
Some models have unverified suites and are omitted from the radar overlay.
DeepSeek V3.2 (Thinking)DeepSeek
80.7VCI
| Metric | DeepSeek V3.2 (Thinking) |
|---|---|
| Vibe Coding Index | 80.7 ✓ |
| SWE-bench Verified | — |
| Terminal-Bench | — |
| LiveCodeBench | 81 ✓ |
| Blended $/M | — |
| Input $/M | — |
| Output $/M | — |
| Output speed (tok/s) | — |
| Latency TTFT (s) | — |
| Context | — |
| Released | — |
| License | Open |