Qwen2.5-VL-72B-Instruct
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
$0.9100
input / 1M
— stable
$0.9100
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=5
Last 30 days
100.0%
n=52
Median response time
4,424ms
n=52
Based on 442 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)
52
OK responses (30d)
52
Total calls (7d)
5
OK responses (7d)
5
Tokonomix benchmark verdicts
Quality dips slightly while latency improves by third
Qwen2.5-VL-72B-Instruct demonstrates a mixed performance shift in this benchmark window. The overall quality score decreased modestly from 99.3 to 98.8, representing a minor regression of 0.5 points. Category performance shows redistribution, with multilingual capabilities improving from 98 to a perfect 100, while coding dropped from 100 to 98. Creative work now scores 98, and reasoning scores are no longer available in the current window data. The most significant improvement comes in latency, which decreased by 32 percent from 9579ms to 6477ms at the median. This represents a substantial gain in responsiveness that users will likely notice in production environments. Despite the small quality decrease, the model maintains exceptionally high performance across all measured categories, remaining above 98 in each area. The vision-capable model continues to deliver strong results for multilingual tasks while maintaining competitive coding and creative output. The latency improvement suggests infrastructure optimizations or model serving enhancements that benefit real-time applications. Users should expect faster responses with marginally adjusted quality characteristics compared to the previous window.
Quality
98.8
Latency p50
6,477 ms
Test runs
5