Side-by-side comparison of pricing, 12 benchmarks, and generation speed.
| Metric | Gemma 4 31B (Reasoning) | Gemini 3 Pro Preview (high) |
|---|---|---|
| Input ($/M tokens) | $0 | $2 |
| Output ($/M tokens) | $0 | $12 |
Verdict. Gemini 3 Pro Preview (high) takes the aggregate benchmark matchup 9–2 across 11 categories. Real workloads usually care about a handful of specific tasks — see the per-benchmark table above.
Pricing. Pricing varies significantly between these models — check the table above for the exact per-token rates. Many production workloads actually surface input-token cost (retrieval-augmented prompts, code-context windows), so factor both directions.
Strengths. Gemma 4 31B (Reasoning) is strongest on GPQA Diamond (86%), IFBench (76%), Coding Index (43.4). Gemini 3 Pro Preview (high) leads on Math Index (95.7), AIME 2025 (96%), LiveCodeBench (92%).
Speed. On throughput, Gemma 4 31B (Reasoning) generates tokens at 36 tok/s versus 0 tok/s — about 100% faster. On time-to-first-token, Gemini 3 Pro Preview (high) responds in 0ms vs 1016ms, which matters most for chat-style UIs.
Provider. Both models come from the same vendor, so the choice comes down to which tier or generation fits your workload — not vendor lock-in.
Workload cost. Workload scenarios (per million requests at 30M input + 15M output tokens): Gemma 4 31B (Reasoning) costs $0.00 ($0/year); Gemini 3 Pro Preview (high) costs $240.00 ($2880/year). At a smaller 5M-input/2M-output scale (single-developer tool or prototype): Gemma 4 31B (Reasoning) ≈ $0.00/run, Gemini 3 Pro Preview (high) ≈ $34.00/run. At agent/realtime scale (200M input / 100M output per million requests): Gemma 4 31B (Reasoning) ≈ $0/run, Gemini 3 Pro Preview (high) ≈ $1600/run. Gemma 4 31B (Reasoning) becomes more attractive at higher volume — the absolute per-token pricing difference compounds when you ship at scale.
Recommendation. Both models have legitimate use cases — the right answer depends on whether you are optimizing for benchmark ceiling, latency, or unit cost. Start with the cheaper / faster model, evaluate against your specific task, and only switch if the upgrade shows a meaningful lift.
Head-to-head deltas
Data from Artificial Analysis API — 12 benchmarks
Gemma 4 31B (Reasoning) is cheaper overall. Its blended price (3:1 input/output ratio) is $0.00/M tokens vs $4.50/M for Gemini 3 Pro Preview (high).
Gemini 3 Pro Preview (high) wins 9 out of 12 benchmarks compared to 2 for Gemma 4 31B (Reasoning). See the detailed benchmark chart above for per-category results.
Gemma 4 31B (Reasoning) generates tokens faster at 36 tok/s vs 0 tok/s. However, Gemini 3 Pro Preview (high) has lower time-to-first-token (0.00s vs 1.02s).
Choose based on your priorities: Gemma 4 31B (Reasoning) for lower cost, Gemini 3 Pro Preview (high) for stronger benchmark performance, and Gemma 4 31B (Reasoning) for faster generation. For latency-sensitive apps, check the TTFT comparison above.