Released: 2025-03-05

QwQ 32B

Alibaba 32B parameter open-weight reasoning model matching frontier model math and coding capabilities.

Intelligence Index
86.1/ 100
Calibrated multi-domain composite
Throughput Speed
45tok/sec
Output streaming throughput
Time To First Token
600ms
Initial chunk server latency
LiveBench Contam-Free
73.8%
September 2026 suite score

LiveBench Multi-Domain Evaluation Breakdown

Monthly refreshed contamination-free test sets
Logical Reasoning86.2%
Coding & Repo Repair74.8%
Mathematics (AIME)89.1%
Data Analysis & Tables75.2%
Language & Comprehension79.5%
Instruction Following85.0%

Sourced Benchmark Evaluations (3)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-2Nearing Saturation67.2%
75
2025-04-02independentSource ↗
SWE-bench VerifiedSaturated56.8%
25
2025-03-14independentSource ↗
IFEvalDeprecated85%
25
2025-03-09independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run qwq-32b

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwq-32b",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for QwQ 32B represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.