Gemini 3.6 Flash and GPT-5 Mini achieved top performance on the seven-day standing benchmark with identical perfect scores of 100%, each requiring $0 in total spend to complete the suite. Gemini 3.1 Flash Lite ranked fourth overall on accuracy at 85.1%, while maintaining a response time under one second during 24-hour latency checks. Across all monitored model endpoints, uptime reached 100% over hourly sampling periods.
Benchmark
The standing benchmark evaluation over the last seven days demonstrates clear performance separation across evaluated models, with zero total spend recorded for all listed runs.
| Model | Standing Benchmark Score | Total Spend |
|---|---|---|
| Gemini 3.6 Flash | 100% | $0 |
| GPT-5 Mini | 100% | $0 |
| GPT-5.6 Terra | 86.8% | $0 |
| Gemini 3.1 Flash Lite | 85.1% | $0 |
| GPT-5.6 Luna | 76% | $0 |
| Gemini 2.5 Flash | 74% | $0 |
Both Gemini 3.6 Flash and GPT-5 Mini captured 100% of available points on the test suite without accruing billed charges. GPT-5.6 Terra followed at 86.8%, outperforming Gemini 3.1 Flash Lite at 85.1%. GPT-5.6 Luna logged 76%, while Gemini 2.5 Flash recorded a score of 74%.
Speed
Latency tracking over the last 24 hours recorded median response times across hourly automated checks. Uptime remained consistent across all systems, with every tested model achieving 100% successful check execution.
Gemini 2.5 Flash Lite posted the fastest response time overall, registering a median latency of 539ms. GPT-5.4 Mini recorded the second fastest speed at 675ms median response time. Two additional models maintained response times below one second: Gemini 2.5 Flash at 906ms median and Gemini 3.1 Flash Lite at 908ms median.
A cluster of models logged median response times between one and two seconds. GPT-5.6 Sol recorded 1.11s, GPT-5.6 Terra reached 1.14s, and GPT-5.6 Luna registered 1.15s. GPT-5 Nano recorded 1.97s median response time.
Slower response times were observed across several high-capacity models. Gemini 3.6 Flash logged 2.02s median response time, followed by GPT-5 Mini at 2.16s and Gemini 3.5 Flash at 2.28s. The longest median latencies were measured on Gemini 2.5 Pro at 3.63s and Gemini 3.1 Pro (preview) at 3.65s.
Prices
Direct API billing observations reflect input and output costs per million tokens across six Google models currently monitored for pricing changes.
Gemini 2.5 Flash Lite presents the lowest baseline operational cost at $0.100 per million input tokens and $0.400 per million output tokens. Gemini 3.1 Flash Lite is priced at $0.250 per million input tokens and $1.50 per million output tokens. Gemini 2.5 Flash carries input pricing of $0.300 per million tokens and output pricing of $2.50 per million tokens.
Higher-tier models exhibit higher per-token rates. Gemini 2.5 Pro costs $1.25 per million input tokens and $10.00 per million output tokens. Gemini 3.5 Flash requires $1.50 per million input tokens alongside $9.00 per million output tokens. Gemini 3.1 Pro (preview) represents the highest pricing tier among monitored models, billed at $2.00 per million input tokens and $12.00 per million output tokens.
These empirical measurements reflect operational conditions and published pricing captured during automated sampling windows. They do not establish performance on unmeasured tasks, guarantee future price stability, or explain internal provider infrastructure decisions.
