In measurements recorded on August 7, 2026, Gemini 3.6 Flash and GPT-5 Mini achieved perfect 100% scores on the 7-day standing benchmark, both requiring $0 in spend. Across all 13 tracked models, API uptime remained at 100% over the last 24 hours of hourly automated checks, though median latency varied significantly from sub-second responses to over four seconds.
Prices
Gemini pricing spans multiple tiers based on model iteration and capability. The lowest billed cost observed belongs to Gemini 2.5 Flash Lite at $0.100 per million input tokens and $0.400 per million output tokens. Gemini 3.1 Flash Lite requires $0.250 per million input tokens and $1.50 per million output tokens, while Gemini 2.5 Flash costs $0.300 per million input tokens and $2.50 per million output tokens.
Higher-tier models scale pricing upward. Gemini 2.5 Pro is billed at $1.25 per million input tokens and $10.00 per million output tokens. Gemini 3.5 Flash sits at $1.50 per million input tokens and $9.00 per million output tokens. The preview release of Gemini 3.1 Pro represents the highest observed cost tier among measured Google models at $2.00 per million input tokens and $12.00 per million output tokens.
Speed
Speed evaluations over the past 24 hours showed a clear split between lightweight models and larger reasoning engines. Gemini 2.5 Flash Lite was the sole model to average under one second, reaching a median response time of 854ms. GPT-5.6 Luna followed at 1.27s, while Gemini 3.1 Flash Lite and Gemini 2.5 Flash tied at 1.45s.
| Model | Median Speed | Uptime |
|---|---|---|
| Gemini 2.5 Flash Lite | 854ms | 100% |
| GPT-5.6 Luna | 1.27s | 100% |
| Gemini 3.1 Flash Lite | 1.45s | 100% |
| Gemini 2.5 Flash | 1.45s | 100% |
| GPT-5.6 Sol | 1.63s | 100% |
| GPT-5.4 Mini | 2.15s | 100% |
| GPT-5.6 Terra | 2.35s | 100% |
| Gemini 3.5 Flash | 2.46s | 100% |
| Gemini 3.6 Flash | 2.90s | 100% |
| GPT-5 Nano | 3.19s | 100% |
| GPT-5 Mini | 3.37s | 100% |
| Gemini 3.1 Pro (preview) | 4.18s | 100% |
| Gemini 2.5 Pro | 4.24s | 100% |
Mid-tier response times included GPT-5.6 Sol at 1.63s, GPT-5.4 Mini at 2.15s, GPT-5.6 Terra at 2.35s, Gemini 3.5 Flash at 2.46s, and Gemini 3.6 Flash at 2.90s. GPT-5 Nano and GPT-5 Mini crossed the three-second threshold at 3.19s and 3.37s, respectively. The slowest models were Gemini 3.1 Pro (preview) at 4.18s and Gemini 2.5 Pro at 4.24s.
Benchmark
The standing benchmark over the last 7 days measures accuracy alongside resource spend. Both Gemini 3.6 Flash and GPT-5 Mini earned 100% of available points with $0 spent. GPT-5.6 Terra followed closely with 91.7% of points at $0 spent. Gemini 2.5 Flash achieved 87.8% of available points while logging $0.00638 in total spend. GPT-5.6 Luna matched that accuracy score at 87.8% with $0 spent. Gemini 3.1 Flash Lite rounded out the evaluated benchmark list at 82.1% of points with $0.00333 spent.
These empirical measurements reflect direct observations under specific synthetic testing conditions during the monitoring period. They demonstrate exact cost, latency, and task accuracy under our standardized methodology, but they do not predict future provider infrastructure adjustments or individual user workload behavior.
