MDL
Qwen3 30B A3B 2507 Instruct
Alibaba
Verified against Artificial Analysis · Jul 13, 2026
Capability
56
percentile index
Price /M
$0.2 / $0.8
in / out
Context
—
Avg score
42.6
11 benchmarks
AA Intelligence
9.1
index
Speed
153 tok/s
TTFT 1s
Released
Jul 29, 2025
About
Qwen3 30B A3B 2507 Instruct is a open-weights model in the Qwen3 family from Alibaba. Benchmarked on 11 evals, averaging 42.6.
Benchmark scores · 11
avg 42.6 — higher is betterThe research behind this model
all papers mentioning it →DeNovoSWE: Scaling Long-Horizon Environments for Generating Entire Repositories from Scratch▲ 34 on HF · Jun 9, 2026MOPD: Multi-Teacher On-Policy Distillation for Capability Integration in LLM Post-Training▲ 15 on HF · Jun 29, 2026Pruning and Distilling Mixture-of-Experts into Dense Language Models▲ 1 on HF · May 27, 2026ConMoE: Expert-Pool Consolidation via Prototype Reassignment for MoE CompressionMay 28, 2026Training-Free Looped TransformersMay 22, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed153 tok/s
Latency (TTFT)1s
Price in / out$0.2 / $0.8 per 1M
Blended price$0.35 / 1M
Math index66.3
Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$0.8/M