Google: Gemini 3.1 Pro Preview vs OpenAI: GPT-5 Codex: Which AI Model Is Better?
Updated March 24, 2026· Based on independent benchmark data
Quick Verdict
Google: Gemini 3.1 Pro Preview leads in intelligence with a score of 57.2 vs 44.6. OpenAI: GPT-5 Codex is 1.6x cheaper at $1.25/1M tokens vs $2.00/1M. For speed, OpenAI: GPT-5 Codex wins at 170 tok/s vs 117 tok/s.
Head-to-Head Comparison
| Metric | Google: Gemini 3.1 Pro Preview | OpenAI: GPT-5 Codex |
|---|---|---|
| Intelligence Score | 57.2 | 44.6 |
| Coding Score | 55.5 | 38.9 |
| Math Score | N/A | 98.7 |
| Speed (tok/s) | 117 tok/s | 170 tok/s |
| Latency (TTFT) | 21.91s | 4.79s |
| Input Price / 1M tokens | $2.00 | $1.25 |
| Output Price / 1M tokens | $12 | $10 |
| Context Window | 1.0M | 400K |
| Max Output Tokens | 66K | 128K |
| Input Modalities | Audio + File + Image + Text + Video | Text + Image |
| Output Modalities | Text | Text |
| Free Tier | No | No |
Detailed Analysis
Intelligence & Quality
Google: Gemini 3.1 Pro Preview outperforms OpenAI: GPT-5 Codex on the Artificial Analysis intelligence index with a score of 57.2 compared to 44.6. For coding tasks, Google: Gemini 3.1 Pro Preview has the edge with a coding score of 55.5 vs 38.9.
Speed & Latency
OpenAI: GPT-5 Codex generates output significantly faster at 170 tok/s compared to Google: Gemini 3.1 Pro Preview's 117 tok/s, making it 1.5x faster for streaming responses. Time to first token is 4.79s for OpenAI: GPT-5 Codex vs 21.91s for Google: Gemini 3.1 Pro Preview, which affects perceived responsiveness in interactive applications.
Pricing
OpenAI: GPT-5 Codex is more affordable at $1.25/1M input tokens ($10/1M output), while Google: Gemini 3.1 Pro Preview costs $2.00/1M input ($12/1M output). For a typical workload of 100 requests per day at 2,000 tokens each, Google: Gemini 3.1 Pro Preview would cost approximately $12.00/month vs $7.50/month for OpenAI: GPT-5 Codex in input costs alone.
Context Window
Google: Gemini 3.1 Pro Preview offers a larger context window at 1.0M tokens compared to OpenAI: GPT-5 Codex's 400K. This means Google: Gemini 3.1 Pro Preview can process roughly 524 pages of text in a single request vs 200 pages for OpenAI: GPT-5 Codex. For output length, OpenAI: GPT-5 Codex can generate up to 128K tokens per response vs 66K for Google: Gemini 3.1 Pro Preview.
Best Use Cases
Choose Google: Gemini 3.1 Pro Preview when you need higher intelligence (57.2), stronger coding performance (55.5), larger context window (1.0M). Choose OpenAI: GPT-5 Codex when you need faster output (170 tok/s), lower cost.
Choose Google: Gemini 3.1 Pro Preview if:
- ✓You need higher intelligence (score: 57.2 vs 44.6)
- ✓You prioritize coding performance (score: 55.5 vs 38.9)
- ✓You need a larger context window (1.0M vs 400K)
Choose OpenAI: GPT-5 Codex if:
- ✓You need faster throughput (170 tok/s vs 117 tok/s)
- ✓You want lower latency (4.79s vs 21.91s TTFT)
- ✓Budget is a concern ($1.25/1M vs $2.00/1M)
Frequently Asked Questions
Is Google: Gemini 3.1 Pro Preview better than OpenAI: GPT-5 Codex for coding?
Google: Gemini 3.1 Pro Preview scores higher on coding benchmarks (55.5 vs 38.9), making it the better choice for programming tasks.
Which is cheaper, Google: Gemini 3.1 Pro Preview or OpenAI: GPT-5 Codex?
OpenAI: GPT-5 Codex is cheaper at $1.25/1M input tokens vs $2.00/1M for Google: Gemini 3.1 Pro Preview.
Is Google: Gemini 3.1 Pro Preview faster than OpenAI: GPT-5 Codex?
OpenAI: GPT-5 Codex is faster, producing output at 170 tok/s compared to Google: Gemini 3.1 Pro Preview's 117 tok/s.
Can Google: Gemini 3.1 Pro Preview process images?
Yes, Google: Gemini 3.1 Pro Preview supports image input. OpenAI: GPT-5 Codex also supports images.
Which has a larger context window, Google: Gemini 3.1 Pro Preview or OpenAI: GPT-5 Codex?
Google: Gemini 3.1 Pro Preview has a larger context window at 1.0M compared to OpenAI: GPT-5 Codex's 400K.
Should I use Google: Gemini 3.1 Pro Preview or OpenAI: GPT-5 Codex?
It depends on your priorities. Google: Gemini 3.1 Pro Preview scores higher on intelligence (57.2), but OpenAI: GPT-5 Codex may be better for specific use cases like budget-conscious projects or speed-critical applications.
Related Comparisons
Benchmark data by Artificial Analysis
Data last synced: March 24, 2026