Google vs 
OpenAI
Verdict
- OpenAI's flagship leads: GPT-6 Astra at 70.6 vs Gemini 3.7 Flash at 64.4.
- OpenAI has the better model in 3 of 4 price bands.
- Google's fastest ranked model (Gemini 2.5 Flash-Lite, 378 tok/s) beats OpenAI's (GPT-5.4 mini, 218 tok/s).
- Google publishes more open-weights models (9 vs 4).
- Google ships more often: one model every 20 days against every 26.
Best model per price band
Each provider’s highest-index model below the price ceiling. Purple marks the winner of the band.
| Price band | OpenAI | |
|---|---|---|
| Under $0.50/M | Gemma 4 26B A4B 54.0$0.190/M | GPT-5.6 Luna xhigh 59.3$0.450/M |
| Under $2/M | Gemini 3.7 Flash medium 64.4$1.50/M | GPT-5.6 Luna xhigh 59.3$0.450/M |
| Under $10/M | Gemini 3.7 Flash medium 64.4$1.50/M | GPT-5.6 Sol high 68.0$8.00/M |
| Any price | Gemini 3.7 Flash medium 64.4$1.50/M | GPT-6 Astra max 70.6$22.00/M |
Line-up
| Metric | OpenAI | |
|---|---|---|
| Flagship index | 64.4 | 70.6 |
| Ranked models | 34 | 38 |
| Open-weights models | 9 | 4 |
| Cheapest ranked, $/M | $0.040 | $0.085 |
| Fastest ranked, tok/s | 378 | 218 |
| Release cadence, days | 20 | 26 |
| Last release | 2 Sept 2026 | 3 Sept 2026 |
| Next expected | 14 Sept 2026 – 30 Sept 2026 | 18 Sept 2026 – 9 Oct 2026 |