Model
Model explorer

Aya Expanse 32B

OPEN
Cohere Labs · Aya Expanse family · released Oct 24, 2024

SOTA-for-size open multilingual 32B model built via data arbitrage, preference tuning, and model merging across 23 languages.

ReasoningCodingVisionFunction callingTool useAgentic
505.2
Elo · rank #311
Parameters
32B
Active params
32B (dense)
Context
128K tokens
Architecture
Dense transformer (Command-based, data arbitrage + model merging), 32B
License
CC-BY-NC
Languages
23+
API price (in/out)
$0.5 / $1.5
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
BIG-Bench HardReasoning56.3%#83
best: ERNIE 4.5 300B-A47B · 94.3%
GPQA DiamondReasoning33.8%#254
best: GPT-6 Astra · 96.0%
IFEvalReasoning68.6%#116
best: Gemma 4 26B A4B · 98.5%
MGSMMath73.4%#32
best: OpenAI o4-mini · 93.7%
MMLU-ProKnowledge45.4%#142
best: Claude Fable 5 · 91.5%
Run it locally
VRAM @ Q4
20 GB
VRAM @ FP16
64 GB
Fits on (Q4)
RTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB
Throughput data unavailable.
Quantizations
GGUF Q4 · EXL2
Fine-tune it
Research only
QLoRA22.2 GB1× RTX 3090 24GB
LoRA68.6 GB1× A100 80GB
Full fine-tune514.0 GB4× H200 141GB
QLoRA SFT on ~10k samples ≈ $16.48 (1× RTX 3090 24GB)
Plan a fine-tune →
Aya Expanse family
Elo progression across releases
Aya Expanse 32BOct 2024505.2
Resources
HFCohereLabs/aya-expanse-32bDOCSdocs.cohere.com/docs/aya-expanseAPIAvailable via Cohere Labs API
API price $0.5/$1.5 · each benchmark row carries its own source badge (see methodology)