head-to-head
| Metric | DeepSeek V4 Pro | Gemini 3.1 Pro |
|---|---|---|
| SWE-bench Verified | 80.6% | 80.6% |
| SWE-bench Pro | 55.4% | 54.2% |
| Terminal-Bench | 67.9% (TB2.0) | — |
| Input $ / 1M | $0.435 | — |
| Context | 1M | — |
| Open weights | Yes | No |
| Maker | DeepSeek | Google DeepMind |
when to pick each
Pick DeepSeek V4 Pro if
The cheapest frontier-class coder — top open-weights score at ~11× less than Opus. Best pick when cost or self-hosting rules.
Pick Gemini 3.1 Pro if
Google's strongest coding model today, with deep Workspace/Cloud integration. (A 3.5 Pro is expected but not shipped.)
Ranked on our AI Coding Leaderboard, updated 2026-07-02. Scores are confirmed against primary sources; prices are per 1M input tokens and can change.
Primary sources
- DeepSeekDeepSeek V4 — specs & benchmarks — Independent tracker (llm-stats, June 2026); tied with Gemini 3.1 Pro on Verified, ahead on Pro.
- Google DeepMindGoogle DeepMind — Gemini Pro — DeepMind-reported pass rate; ties DeepSeek V4 on Verified, trails it on Pro.
- BenchmarkSWE-bench — the real-GitHub-issue benchmark