EXAONE 4.0 1.2B

LG AI Research · lgai-exaone/exaone-4-0-1-2b

ProviderLG AI Research
Familyexaone
Typellm-chat
Statusactive
Released2025-07-11
Updated2025-08-04
Parameters1.3B
Open weightsyes

What it fits on

Ordered by predicted speed. Bandwidth sets decode rate; memory decides whether it runs at all.

Every figure here is computed, not measured. Fit is weights at each quantisation against device memory, with a 25% allowance for the KV cache, activations and the OS. Decode rate is the memory-bandwidth roofline at 70% efficiency. Nobody has run this model on these devices.
DeviceMemoryBandwidthBest qualityWeightstok/sFastest
Cerebras WSE-344.0 GB21000000.0 GB/sbf162.56 GB~5744918.6~22979674.5 q4
NVIDIA GB200 Grace Blackwell Superchip372.0 GB16000.0 GB/sbf162.56 GB~4377.1~17508.3 q4
AMD Instinct MI355X288.0 GB8000.0 GB/sbf162.56 GB~2188.5~8754.2 q4
NVIDIA B200180.0 GB7700.0 GB/sbf162.56 GB~2106.5~8425.9 q4
Google TPU7x (Ironwood)192.0 GB7380.0 GB/sbf162.56 GB~2018.9~8075.7 q4
AMD Instinct MI325X256.0 GB6000.0 GB/sbf162.56 GB~1641.4~6565.6 q4
AMD Instinct MI300X192.0 GB5300.0 GB/sbf162.56 GB~1449.9~5799.6 q4
NVIDIA H200 SXM141.0 GB4800.0 GB/sbf162.56 GB~1313.1~5252.5 q4
NVIDIA H100 NVL94.0 GB3900.0 GB/sbf162.56 GB~1066.9~4267.7 q4
NVIDIA H100 SXM80.0 GB3350.0 GB/sbf162.56 GB~916.5~3665.8 q4
AMD Instinct MI250X128.0 GB3200.0 GB/sbf162.56 GB~875.4~3501.7 q4
Google TPU v5p95.0 GB2765.0 GB/sbf162.56 GB~756.4~3025.7 q4
NVIDIA A100 80GB SXM80.0 GB2039.0 GB/sbf162.56 GB~557.8~2231.2 q4
NVIDIA H100 PCIe80.0 GB2000.0 GB/sbf162.56 GB~547.1~2188.5 q4
NVIDIA GeForce RTX 509032.0 GB1792.0 GB/sbf162.56 GB~490.2~1960.9 q4
Google TPU v6e (Trillium)32.0 GB1638.0 GB/sbf162.56 GB~448.1~1792.4 q4
AMD Instinct MI21064.0 GB1600.0 GB/sbf162.56 GB~437.7~1750.8 q4
NVIDIA A100 40GB SXM40.0 GB1555.0 GB/sbf162.56 GB~425.4~1701.6 q4
Google TPU v432.0 GB1200.0 GB/sbf162.56 GB~328.3~1313.1 q4
NVIDIA GeForce RTX 409024.0 GB1008.0 GB/sbf162.56 GB~275.8~1103.0 q4
AMD Radeon RX 7900 XTX24.0 GB960.0 GB/sbf162.56 GB~262.6~1050.5 q4
NVIDIA GeForce RTX 508016.0 GB960.0 GB/sbf162.56 GB~262.6~1050.5 q4
NVIDIA GeForce RTX 309024.0 GB936.0 GB/sbf162.56 GB~256.1~1024.2 q4
NVIDIA L40S48.0 GB864.0 GB/sbf162.56 GB~236.4~945.4 q4
Apple M3 Ultra512.0 GB819.0 GB/sbf162.56 GB~224.1~896.2 q4
AMD Radeon RX 7900 XT20.0 GB800.0 GB/sbf162.56 GB~218.9~875.4 q4
Apple M2 Ultra192.0 GB800.0 GB/sbf162.56 GB~218.9~875.4 q4
Google TPU v5e16.0 GB800.0 GB/sbf162.56 GB~218.9~875.4 q4
NVIDIA GeForce RTX 4080 SUPER16.0 GB736.0 GB/sbf162.56 GB~201.3~805.4 q4
NVIDIA GeForce RTX 4070 Ti SUPER16.0 GB672.0 GB/sbf162.56 GB~183.8~735.3 q4
AMD Radeon RX 9070 XT16.0 GB640.0 GB/sbf162.56 GB~175.1~700.3 q4
Apple M4 Max128.0 GB546.0 GB/sbf162.56 GB~149.4~597.5 q4
Apple M1 Max64.0 GB400.0 GB/sbf162.56 GB~109.4~437.7 q4
Apple M2 Max96.0 GB400.0 GB/sbf162.56 GB~109.4~437.7 q4
Apple M3 Max128.0 GB400.0 GB/sbf162.56 GB~109.4~437.7 q4
NVIDIA GeForce RTX 3060 12GB12.0 GB360.0 GB/sbf162.56 GB~98.5~393.9 q4
NVIDIA L424.0 GB300.0 GB/sbf162.56 GB~82.1~328.3 q4
NVIDIA GeForce RTX 4060 Ti 16GB16.0 GB288.0 GB/sbf162.56 GB~78.8~315.1 q4
Apple M4 Pro64.0 GB273.0 GB/sbf162.56 GB~74.7~298.7 q4
Apple M432.0 GB120.0 GB/sbf162.56 GB~32.8~131.3 q4

Where it runs

Taken from the card's availability section.

PlatformId on that platform
Hugging FaceLGAI-EXAONE/EXAONE-4.0-1.2B

Verified benchmark evidence

Each row was checked against its source by a reviewer, and carries the model identifier as actually evaluated — which is not always the same as this card's.
BenchmarkEvaluated asScoreEvidence dateSource kind
aa_lcrExaone 4.0 1.2B0.0%2026-09-10 evaluatedindependent evaluatorsource
critptExaone 4.0 1.2B0.0%2026-09-10 evaluatedindependent evaluatorsource
gpqa_diamondExaone 4.0 1.2B51.52%2026-09-10 evaluatedindependent evaluatorsource

Reported benchmark scores

This card reports no benchmark scores yet.

BenchmarkCatalogue standingScore

Data

This card as JSON · See it in the graph · Edit on GitHub