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
Evaluation results from judge-model scoring across diverse task categories. Scores reflect coherence, accuracy and instruction-following.
Pricing history
Direct provider rates per million tokens, plus a typical-conversation cost estimate.
Pricing over time
Input & output per 1M tokens · step-line = price changes
$0.0900
input / 1M
— stable
$0.2800
output / 1M
— stable
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=1,231
Last 30 days
100.0%
n=6,099
Median response time
1,915ms
n=6,099
Based on 6,479 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)
6,099
OK responses (30d)
6,099
Total calls (7d)
1,231
OK responses (7d)
1,231
Tokonomix benchmark verdicts
Quality drops 10.2 points to 84.6 amid 38% latency increase
Mistral-Small-3.2-24B-Instruct-2506 experienced notable performance degradation in this benchmark window, with overall quality declining from 94.8 to 84.6 points while latency increased by 38% to a median of 6559 milliseconds. The model maintained exceptional multilingual capabilities at 100 points, consistent with previous performance. However, significant shifts occurred in tested categories: coding performance disappeared from evaluation while new reasoning scores emerged strong at 95 points. Creative output remained relatively stable, moving from 85 to 87 points. The most concerning change appears in factual accuracy, which scored only 57 points in the current window, representing a substantial weakness compared to the model's other capabilities. The combination of slower response times and lower quality scores suggests possible infrastructure or configuration issues at the OVH AI Endpoints GRA deployment. Users should expect longer wait times for responses and exercise caution with factual queries, though the model continues to excel at multilingual tasks and demonstrates strong reasoning abilities. The performance decline warrants monitoring in upcoming benchmark windows to determine whether this represents a temporary regression or a sustained shift in model behavior.
Quality
84.6
Latency p50
6,559 ms
Test runs
5