Speed analysis
Latency measured across all benchmark runs. P50 (median) and P95 (95th percentile) give a realistic picture of response speed under normal and peak load.
Quality scores
How this model compares to the rest of the field on each prompt category, from a pairwise fit over the same prompts. The raw judge score sits underneath each number.
Win rate per category: how often this model beats a field-average model on a prompt from that category. 50% is average, not a failing grade. It is not a percentage of correct answers.
Pricing
What you pay per million tokens when you use this model on Tokonomix, plus an estimate for a typical conversation.
Tokens per second
Throughput in tokens per second, derived from measured P50 latency. Higher is better; fluctuations track provider-side load.
Estimated from P50 latency × 200 output tokens — the absolute number depends on this assumption; the trend is what matters.
Capabilities
Availability
Availability
How often this model answers when we call it — measured across real API requests and live tests over the last 30 days. This is separate from quality: these numbers only tell you whether the model responds, not how good the answer is.
Last 7 days
100.0%
n=19
Last 30 days
100.0%
n=22
Median response time
4,604ms
n=22
Based on 407 measurements over the last 30 days.
Technical details
Only live API calls and live-test requests count — internal probes and benchmark runs are excluded.
Calls with a custom API key (BYOK) are excluded: those failures are key-specific, not a sign of model downtime.
Failed calls are NOT included in quality scores — quality is measured on successful responses only. Availability and quality are independent signals.
Median response time (p50) across successful calls with a recorded duration. Outliers (very slow or very fast calls) pull the median less than the average.
Total calls (30d)
22
OK responses (30d)
22
Total calls (7d)
19
OK responses (7d)
19
Tokonomix benchmark verdicts
Quality drops 13.3 points to 78.7 amid latency and scoring shifts
Meta-Llama-3_3-70B-Instruct shows notable performance degradation in this benchmark window, with the overall quality score falling from 92.0 to 78.7. The median latency increased by 45 percent, rising from 8405 ms to 12224 ms, indicating slower response times that may impact user experience. Category performance reveals mixed results with reasoning maintaining exceptional performance at 100, while coding capability declined from 92 to 83. The factual accuracy score of 53 represents a concerning weakness, though this metric was not measured in the previous window. Previously strong multilingual and creative scores are absent from current testing, making direct comparison challenging. The model continues to demonstrate solid reasoning capabilities and acceptable coding performance, but the substantial drop in overall quality combined with increased latency suggests potential infrastructure or configuration changes. Users should expect longer wait times and may encounter reduced accuracy in factual tasks. The reasoning category remains a strength worth noting, but the overall trajectory indicates this benchmark window represents a step backward for the model's performance on the OVH AI Endpoints platform.
Quality
78.7
Latency p50
12,224 ms
Test runs
5