Private AI
Qwen3.5 122B A10B is a large open Qwen 3.5 MoE model served through Gerra with tool calling and structured output support.
Added Feb 24, 2026
Model weightsContext Window
131.1K
Max Output
32.8K
Input Price (Auto)
$0.46/1M
Output Price (Auto)
$3.67/1M
Cache Read (Auto)
$0.11/1M
Capabilities
Performance metrics and benchmarks
Sourced from Artificial Analysis.
Intelligence Index
27.6
Choose explicit providers for this model. Auto routing remains available as the default option.
Loading provider options…
Coding Index
43.3
Agentic Index
15.8
GPQA Diamond
Graduate-level scientific reasoning
82.7%
Better than 76% of models compared
HLE
Humanity's Last Exam
14.8%
Better than 72% of models compared
IFBench
Instruction-following benchmark
50.8%
Better than 60% of models compared
T²-Bench Telecom
Conversational AI agents in dual-control scenarios
84.5%
Better than 75% of models compared
AA-LCR
Long context reasoning evaluation
56.0%
Better than 65% of models compared
GDPval-AA
Economically valuable tasks
19.4%
CritPt
Research-level physics reasoning
0.9%
SciCode
Python programming for scientific computing
35.6%
Better than 55% of models compared
Terminal-Bench Hard
Agentic coding and terminal use
29.5%
AA-Omniscience Accuracy
Proportion of correctly answered questions
18.6%
AA-Omniscience Hallucination Rate
Rate of incorrect answers among non-correct responses
89.1%
Last updated Aug 2, 2026, 12:00 PM
Artificial AnalysisBetter than 71% of models compared
