In telemetry collected on September 2, 2026, Gemini 3.6 Flash and GPT-5 Mini achieved identical 100% completion rates on the standing 7-day benchmark, both recording $0 in total evaluation spend. Meanwhile, latency tests over the past 24 hours revealed a wide performance spread across models, with [Gemini 2.5 Flash Lite](/models/google-gemini-2-5-flash-lite) registering the fastest response time at 577ms and Gemini 2.5 Pro trailing at 3.71s median response time.
Speed
All 13 evaluated model endpoints maintained 100% uptime across hourly checks over the last 24 hours. Lightweight models dominated the top latency tiers. Gemini 2.5 Flash Lite led all systems with a 577ms median response time, followed by GPT-5.4 Mini at 677ms. Gemini 3.1 Flash Lite and Gemini 2.5 Flash posted nearly identical speeds at 897ms and 898ms median response times, respectively.
Mid-range models clustered in the 1-second to 2-second range. GPT-5.6 Terra recorded a 1.14s median response time, GPT-5.6 Luna reached 1.19s, and GPT-5.6 Sol registered 1.67s. Larger Flash variants required more processing time, with Gemini 3.5 Flash logging 2.41s median latency and Gemini 3.6 Flash recording 2.54s. OpenAI's compact offerings showed similar timing, as GPT-5 Nano hit 2.61s median latency while GPT-5 Mini reached 2.63s. Pro-tier models logged the slowest response times in the 24-hour evaluation window, led by Gemini 3.1 Pro (preview) at 3.62s and Gemini 2.5 Pro at 3.71s.
| Model | Median Response Time | Uptime |
|---|---|---|
| Gemini 2.5 Flash Lite | 577ms | 100% |
| GPT-5.4 Mini | 677ms | 100% |
| Gemini 3.1 Flash Lite | 897ms | 100% |
| Gemini 2.5 Flash | 898ms | 100% |
| GPT-5.6 Terra | 1.14s | 100% |
| GPT-5.6 Luna | 1.19s | 100% |
| GPT-5.6 Sol | 1.67s | 100% |
| Gemini 3.5 Flash | 2.41s | 100% |
| Gemini 3.6 Flash | 2.54s | 100% |
| GPT-5 Nano | 2.61s | 100% |
| GPT-5 Mini | 2.63s | 100% |
| Gemini 3.1 Pro (preview) | 3.62s | 100% |
| Gemini 2.5 Pro | 3.71s | 100% |
Benchmark
Standing benchmark evaluations across the last 7 days highlighted accuracy contrasts among the six tracked models, with all evaluations incurring $0 in spend. Gemini 3.6 Flash and GPT-5 Mini tied for the top position, earning 100% of available points. GPT-5.6 Terra placed third with 89.9%, followed by Gemini 3.1 Flash Lite at 87.5% and Gemini 2.5 Flash at 83.3%. GPT-5.6 Luna recorded the lowest score in the set at 69.8%.
Prices
Billed API costs across Google's Gemini models reflect distinct pricing tiers per million tokens. Gemini 2.5 Flash Lite remains the lowest-cost option in the matrix at $0.100 per million input tokens and $0.400 per million output tokens. Gemini 3.1 Flash Lite costs $0.250 per million input tokens and $1.50 per million output tokens, while Gemini 2.5 Flash is priced at $0.300 per million input tokens and $2.50 per million output tokens.
Higher-tier models scale up significantly in cost. Gemini 2.5 Pro is billed at $1.25 per million input tokens and $10.00 per million output tokens. Gemini 3.5 Flash costs $1.50 per million input tokens and $9.00 per million output tokens. Gemini 3.1 Pro (preview) sits at the top of the price table at $2.00 per million input tokens and $12.00 per million output tokens.
These measurements confirm operational availability and benchmark rates under observed conditions during the test window. They do not predict long-term stability, real-world application throughput, or future pricing adjustments by API providers.
