Qwen3.5-397B-A17B
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.
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
Tokonomix benchmark verdicts
Qwen3.5-397B-A17B establishes baseline with strong creative performance
This first benchmark window establishes baseline performance for Qwen3.5-397B-A17B deployed through OVH AI Endpoints in the GRA region. The model demonstrates particularly strong creative writing capabilities, achieving 9.0 out of 10 in creative tasks, indicating robust narrative generation and imaginative content production. Coding performance is solid at 7.5, showing competence in programming tasks though with room for optimization. Mathematical reasoning scores 7.0, representing adequate performance for standard computational problems. The model handles instruction following reliably at 7.0, meeting basic compliance requirements. Response coherence is maintained at 7.0, ensuring outputs remain logical and well-structured. Overall performance across all categories averages a respectable level for a model of this class. Users should expect best results when leveraging the model for creative content generation, storytelling, and narrative tasks. For production code generation and complex mathematical proofs, outputs may require additional validation. This baseline provides a reference point for tracking future performance trends and model updates.
Quality
—
Latency p50
—
Test runs
0