Qwen 3 235B vs Llama 4 Maverick
Architecture Comparison
SpecQwen 3 235BLlama 4 Maverick
TypeMOEMOE
Total Parameters235B400B
Active Parameters22B17B
Layers9496
Hidden Dimension5,1205,120
Attention Heads6440
KV Heads48
Context Length131,0721,048,576
Precision (default)BF16BF16
Total Experts128128
Active Experts81
Memory Requirements
PrecisionQwen 3 235BLlama 4 Maverick
BF16 Weights470.0 GB800.0 GB
FP8 Weights235.0 GB400.0 GB
INT4 Weights117.5 GB200.0 GB
KV-Cache / Token192512 B393216 B
Activation Estimate3.00 GB3.00 GB
Minimum GPUs Needed (BF16)
H100 SXM7 GPUsN/A
L40SN/AN/A
Quality Benchmarks
BenchmarkQwen 3 235BLlama 4 Maverick
Overall8889
MMLU88.089.0
HumanEval62.063.0
GSM8K94.095.0
MT-Bench88.088.0
Qwen 3 235B
MMLU
88.0
HumanEval
62.0
GSM8K
94.0
MT-Bench
88.0
Llama 4 Maverick
MMLU
89.0
HumanEval
63.0
GSM8K
95.0
MT-Bench
88.0
Capabilities
FeatureQwen 3 235BLlama 4 Maverick
Tool Use✓ Yes✓ Yes
Vision✗ No✓ Yes
Code✓ Yes✓ Yes
Math✓ Yes✓ Yes
Reasoning✓ Yes✗ No
Multilingual✓ Yes✓ Yes
Structured Output✓ Yes✓ Yes
API Pricing Comparison
Cheapest Output (Qwen 3 235B)
$3.00/M
Input: $1.50/M
Cheapest Output (Llama 4 Maverick)
$1.80/M
Input: $1.20/M
| Provider | Qwen 3 235B In $/M | Out $/M | Llama 4 Maverick In $/M | Out $/M |
|---|---|---|---|---|
| together | $1.50 | $3.00 | $1.20 | $1.80 |
| fireworks | $1.80 | $3.50 | $1.50 | $2.00 |
Recommendation Summary
- ‣Llama 4 Maverick scores higher on overall quality (89 vs 88).
- ‣Llama 4 Maverick is cheaper per output token ($1.80/M vs $3.00/M).
- ‣Qwen 3 235B has a smaller memory footprint (470.0 GB vs 800.0 GB BF16), making it easier to deploy on fewer GPUs.
- ‣Llama 4 Maverick supports a longer context window (1,048,576 vs 131,072 tokens).
- ‣Llama 4 Maverick is stronger at code generation (HumanEval: 63.0 vs 62.0).
- ‣Llama 4 Maverick is better at math reasoning (GSM8K: 95.0 vs 94.0).