Model Details
Qwen3 Next 80B A3B Instruct
Quick answer: Qwen3 Next 80B A3B Instruct is Alibaba's April 2026 MoE model scoring 82.7% Arena Hard v2, 80.6% MMLU-Pro, and 70.3% BFCL-v3. Apache 2.0 — 80B total / 3B active.
At a Glance
Where Qwen3 Next 80B A3B Instruct leads
- 82.7% Arena Hard v2 — excellent instruction quality
- 80.6% MMLU-Pro — strong academic knowledge
- 70.3% BFCL-v3 — solid function calling
- Apache 2.0 — fully open
- 3B active: low inference cost
Where it lags
- Below Qwen3 Next 80B A3B Thinking on MMLU-Pro (80.6% vs 82.7%)
Best for: High-instruction-quality open-source deployments; Apache 2.0 Arena Hard benchmark pipelines at 3B active cost.
What Qwen3 Next 80B A3B Instruct Is
Qwen3 Next 80B A3B Instruct is the April 2026 instruct variant in the Qwen3 Next generation. The 82.7% Arena Hard is notably high for an Apache 2.0 MoE model.
Specifications
| Field | Value |
|---|---|
| Organization | Alibaba |
| License | Apache 2.0 |
| HuggingFace | Qwen/Qwen3-Next-80B-A3B-Instruct |
| Release date | April 2026 |
| Parameters | 80B total / 3B active (MoE) |
| Modality | Text only |
Pricing
Open weights under Apache 2.0 — self-host at no cost.
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| Arena Hard v2 | 82.7% | Benchgen evaluation | 2026-04 |
| MMLU-Pro | 80.6% | Benchgen evaluation | 2026-04 |
| BFCL-v3 | 70.3% | Benchgen evaluation | 2026-04 |
Qwen3 Next 80B Instruct vs Alternatives
| Model | Arena Hard | MMLU-Pro | BFCL-v3 | Active Params |
|---|---|---|---|---|
| Qwen3 Next 80B A3B Instruct | 82.7% | 80.6% | 70.3% | 3B |
| Qwen3 Next 80B A3B Thinking | 62.3% | 82.7% | 72.0% | 3B |
| Mistral Small 3 24B | 87.6% | 66.3% | — | 24B dense |
Qwen3 Next 80B Instruct leads on Arena Hard (82.7% vs 62.3% Thinking). For maximum MMLU-Pro at 3B active: use Thinking (82.7%).
Related Models & Benchmarks
- Compare: Qwen3 Next 80B A3B Thinking · Qwen3.6 35B A3B
- Benchmarks: Arena Hard v2 · MMLU-Pro · BFCL-v3
Frequently Asked Questions
What is Qwen3 Next 80B A3B Instruct?
Alibaba's April 2026 MoE instruct model scoring 82.7% Arena Hard, 80.6% MMLU-Pro. Apache 2.0 — 80B/3B active.Specs from Alibaba's Qwen3 Next 80B A3B Instruct release (April 2026) and Benchgen evaluations. Last updated 2026-07-24.
Benchmark Leaderboards
This model isn’t on any benchmark leaderboard yet.