Back to Explore
Qwen: Qwen3 Next 80B A3B Instruct
Alibaba·Released 2025-09-11
Open Source80B262K ctxApache 2.0
About
Qwen3-Next-80B-A3B-Instruct is an instruction-tuned chat model in the Qwen3-Next series optimized for fast, stable responses without “thinking” traces. It targets complex tasks across reasoning, code generation, knowledge QA, and multilingual...
Related Models
Pricing
Input
$0.09
per 1M tokens
Output
$1.10
per 1M tokens
Blended
$0.34
per 1M tokens
Cheaper than 60% of models. Median price is $0.54/1M tokens.
Cost Calculator
Tokens per day1M
100K100M
Daily
$0.34
Monthly
$10.28
vs. Similar Models
Qwen: Qwen3 Coder 30B A3B InstructQ:-0.1
$0.12-65%
QwQ 32BQ:-0.3
$0.74+118%
Qwen3 235B A22B (Reasoning)Q:-0.3
$2.63+666%
Qwen3 VL 30B A3B (Reasoning)Q:-0.4
$0.34-1%
Performance
189
tokens/sec
Faster than 85% of models
1.11
seconds
Faster than 50% of models
1.11
seconds
Faster than 66% of models
Market Median
94 tok/s
100% faster
Median TTFT
1.11s
1% faster
Throughput/Dollar
551
tok/s per $/1M
Speed Comparison
Gemini 3 Flash Preview (Non-reasoning)
188 tok/s-0%
Step 3.5 Flash
189 tok/s+0%
Nova 2.0 Lite (high)
188 tok/s-0%
Context Window
262K
tokens
Larger than 62% of models
Max Output
16K
tokens
6% of context
Benchmarks
MMLU-Pro
81.9%
GPQA Diamond
73.8%
HLE
7.3%
LiveCodeBench
68.4%
SciCode
30.7%
TerminalBench Hard
7.6%
MATH-500Not evaluated
AIMENot evaluated
AIME 2025
66.3%
IFBench
39.7%
Long Context Recall
51.3%
Tau2
21.6%
Market AverageTop Score
Open Source
apache-2.080BGGUF / GPTQ / AWQ
Downloads
777.1K
Likes
951
VRAM (FP16)
Multi-GPU
GPU
8x A100 / H100