MDL
Qwen3 Next 80B A3B Instruct
Alibaba
Verified against Artificial Analysis · Jul 13, 2026
Capability
61
percentile index
Price /M
$0.5 / $2
in / out
Context
262.1K
tokens
Avg score
44.9
10 benchmarks
AA Intelligence
13.7
index
Speed
189 tok/s
TTFT 1s
Released
Sep 11, 2025
About
Qwen3 Next 80B A3B Instruct is a open-weights model in the Qwen3 family from Alibaba. Benchmarked on 10 evals, averaging 44.9.
Benchmark scores · 10
avg 44.9 — higher is betterThe research behind this model
all papers mentioning it →KVBuffer: IO-aware Serving for Linear AttentionMay 18, 2026Provably Shorter Scratchpads in Hybrid DeltaNet-Attention DecodersMay 15, 2026CRANE: Constrained Reasoning Injection for Code Agents via Nullspace EditingMay 13, 2026SlimQwen: Exploring the Pruning and Distillation in Large MoE Model Pre-trainingMay 9, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed189 tok/s
Latency (TTFT)1s
Price in / out$0.5 / $2 per 1M
Blended price$0.875 / 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$2/M