Qwen 1.5 MoE A2.7B
Alibaba · moe · 14.3B parameters · 32,768 context
Parameters
14.3B
Context Window
32K tokens
Architecture
MoE
Best GPU
A100 40GB SXM
Intelligence Brief
Qwen 1.5 MoE A2.7B is a 14.3B parameter Mixture-of-Experts (60 experts, 4 active) model from Alibaba, featuring Multi-Head Attention (MHA) with 24 layers and 2,048 hidden dimensions. With a 32,768 token context window, it supports structured output, code, math, multilingual. For self-hosted inference, A100 40GB SXM delivers optimal throughput at $807/month.
Recent changes
Loading…
Related models
5 suggestions- Qwen 2.5 3BQwen 2.5 · 3.09B$0.100/M out
- Qwen 2 VL 2BQwen 2 VL · 2.2B—
- Qwen 2.5 1.5BQwen 2.5 · 1.5B—
- Qwen 2.5 Coder 1.5BQwen 2.5 Coder · 1.5B—
- Qwen 2.5 Coder 3BQwen 2.5 Coder · 3.1B—
Picks: same family first, then same vendor within ±2× params, then top tag-overlap matches. Price shown is the cheapest Output $/M across providers — the row's page shows the canonical anchor.
Architecture Details
Memory Requirements
BF16 Weights
28.6 GB
FP8 Weights
14.3 GB
INT4 Weights
7.2 GB
GPU Compatibility Matrix
Qwen 1.5 MoE A2.7B is compatible with 82% of GPU configurations across 41 GPUs at 3 precision levels.
GPU Recommendations
BF16 · 1 GPU · vllm
100/100
score
Throughput
1.1K tok/s
Latency (ITL)
1.0ms
Est. TTFT
0ms
Cost/Month
$807
Cost/M Tokens
$0.29
BF16 · 1 GPU · vllm
100/100
score
Throughput
768.0 tok/s
Latency (ITL)
1.3ms
Est. TTFT
0ms
Cost/Month
$465
Cost/M Tokens
$0.23
BF16 · 1 GPU · vllm
100/100
score
Throughput
696.0 tok/s
Latency (ITL)
1.4ms
Est. TTFT
0ms
Cost/Month
$399
Cost/M Tokens
$0.22
Deployment Options
API Deployment
No API pricing available
Single GPU
A100 40GB SXM
$807/mo
Min VRAM: 14 GB
Multi-GPU
A10G x2
813.3 tok/s
TP· $569/mo
API Pricing Comparison
No API pricing data available for this model.
Performance Estimates
Throughput by GPU
VRAM Breakdown (A100 40GB SXM, BF16)
Precision Impact
bf16
28.6 GB
weights/GPU
~1.1K tok/s
fp8
14.3 GB
weights/GPU
int4
7.2 GB
weights/GPU
Capabilities
Features
Supported Frameworks
Supported Precisions
Where to Deploy Qwen 1.5 MoE A2.7B
Similar Models
RWKV-6 14B
14.1B params · hybrid
Quality: 50
from $0.20/M
Phi 3 Medium 14B
14B params · dense
Quality: 76
Nekomata 14B
14B params · dense
Quality: 50
Phi-4
14.7B params · dense
Quality: 73
from $0.14/M
Qwen 2.5 Coder 14B
14.7B params · dense
Quality: 50
from $0.30/M
Frequently Asked Questions
How much VRAM does Qwen 1.5 MoE A2.7B need for inference?
Qwen 1.5 MoE A2.7B requires approximately 28.6 GB of VRAM at BF16 precision, 14.3 GB at FP8, or 7.2 GB at INT4 quantization. Additional VRAM is needed for KV-cache (196608 bytes per token) and activations (~0.50 GB).
What is the best GPU for Qwen 1.5 MoE A2.7B?
The top recommended GPU for Qwen 1.5 MoE A2.7B is the A100 40GB SXM using BF16 precision. It achieves approximately 1.1K tokens/sec at an estimated cost of $807/month ($0.29/M tokens). Score: 100/100.
How much does Qwen 1.5 MoE A2.7B inference cost?
Qwen 1.5 MoE A2.7B inference costs vary by provider and GPU setup. Use our calculator for detailed cost estimates across all providers.