Motif-2-12.7B

Motif Technologies · motif/motif-2-12-7b-reasoning

ProviderMotif Technologies
Familymotif
Typellm-reasoning
Statusactive
Released2025-11-28
Updated2026-03-10
Parameters12.7B
Open weightsyes

Lineage

RelationshipModel
Is a finetune ofMotif-Technologies/Motif-2-12.7B-Base

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/sbf1625.41 GB~578564.3~2314257.1 q4
NVIDIA GB200 Grace Blackwell Superchip372.0 GB16000.0 GB/sbf1625.41 GB~440.8~1763.2 q4
AMD Instinct MI355X288.0 GB8000.0 GB/sbf1625.41 GB~220.4~881.6 q4
NVIDIA B200180.0 GB7700.0 GB/sbf1625.41 GB~212.1~848.6 q4
Google TPU7x (Ironwood)192.0 GB7380.0 GB/sbf1625.41 GB~203.3~813.3 q4
AMD Instinct MI325X256.0 GB6000.0 GB/sbf1625.41 GB~165.3~661.2 q4
AMD Instinct MI300X192.0 GB5300.0 GB/sbf1625.41 GB~146.0~584.1 q4
NVIDIA H200 SXM141.0 GB4800.0 GB/sbf1625.41 GB~132.2~529.0 q4
NVIDIA H100 NVL94.0 GB3900.0 GB/sbf1625.41 GB~107.4~429.8 q4
NVIDIA H100 SXM80.0 GB3350.0 GB/sbf1625.41 GB~92.3~369.2 q4
AMD Instinct MI250X128.0 GB3200.0 GB/sbf1625.41 GB~88.2~352.6 q4
Google TPU v5p95.0 GB2765.0 GB/sbf1625.41 GB~76.2~304.7 q4
NVIDIA A100 80GB SXM80.0 GB2039.0 GB/sbf1625.41 GB~56.2~224.7 q4
NVIDIA H100 PCIe80.0 GB2000.0 GB/sbf1625.41 GB~55.1~220.4 q4
NVIDIA GeForce RTX 509032.0 GB1792.0 GB/sfp812.7 GB~98.7~197.5 q4
Google TPU v6e (Trillium)32.0 GB1638.0 GB/sfp812.7 GB~90.3~180.5 q4
AMD Instinct MI21064.0 GB1600.0 GB/sbf1625.41 GB~44.1~176.3 q4
NVIDIA A100 40GB SXM40.0 GB1555.0 GB/sbf1625.41 GB~42.8~171.4 q4
Google TPU v432.0 GB1200.0 GB/sfp812.7 GB~66.1~132.2 q4
NVIDIA GeForce RTX 409024.0 GB1008.0 GB/sfp812.7 GB~55.5~111.1 q4
AMD Radeon RX 7900 XTX24.0 GB960.0 GB/sfp812.7 GB~52.9~105.8 q4
NVIDIA GeForce RTX 508016.0 GB960.0 GB/sq69.53 GB~70.5~105.8 q4
NVIDIA GeForce RTX 309024.0 GB936.0 GB/sfp812.7 GB~51.6~103.1 q4
NVIDIA L40S48.0 GB864.0 GB/sbf1625.41 GB~23.8~95.2 q4
Apple M3 Ultra512.0 GB819.0 GB/sbf1625.41 GB~22.6~90.3 q4
AMD Radeon RX 7900 XT20.0 GB800.0 GB/sfp812.7 GB~44.1~88.2 q4
Apple M2 Ultra192.0 GB800.0 GB/sbf1625.41 GB~22.0~88.2 q4
Google TPU v5e16.0 GB800.0 GB/sq69.53 GB~58.8~88.2 q4
NVIDIA GeForce RTX 4080 SUPER16.0 GB736.0 GB/sq69.53 GB~54.1~81.1 q4
NVIDIA GeForce RTX 4070 Ti SUPER16.0 GB672.0 GB/sq69.53 GB~49.4~74.1 q4
AMD Radeon RX 9070 XT16.0 GB640.0 GB/sq69.53 GB~47.0~70.5 q4
Apple M4 Max128.0 GB546.0 GB/sbf1625.41 GB~15.0~60.2 q4
Apple M1 Max64.0 GB400.0 GB/sbf1625.41 GB~11.0~44.1 q4
Apple M2 Max96.0 GB400.0 GB/sbf1625.41 GB~11.0~44.1 q4
Apple M3 Max128.0 GB400.0 GB/sbf1625.41 GB~11.0~44.1 q4
NVIDIA GeForce RTX 3060 12GB12.0 GB360.0 GB/sq57.94 GB~31.7~39.7 q4
NVIDIA L424.0 GB300.0 GB/sfp812.7 GB~16.5~33.1 q4
NVIDIA GeForce RTX 4060 Ti 16GB16.0 GB288.0 GB/sq69.53 GB~21.2~31.7 q4
Apple M4 Pro64.0 GB273.0 GB/sbf1625.41 GB~7.5~30.1 q4
Apple M432.0 GB120.0 GB/sfp812.7 GB~6.6~13.2 q4

Where it runs

Taken from the card's availability section.

PlatformId on that platform
Hugging FaceMotif-Technologies/Motif-2-12.7B-Reasoning

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_lcrMotif-2-12.7B14.67%2026-09-10 evaluatedindependent evaluatorsource
critptMotif-2-12.7B0.0%2026-09-10 evaluatedindependent evaluatorsource
gpqa_diamondMotif-2-12.7B69.49%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