Ling 3.0 Tiny

inclusionAI · inclusionai/ling-3-0-tiny

ProviderinclusionAI
Familyling
Typellm-chat
Statusactive
Released2026-08-10
Updated2026-09-04
Parameters7.9B
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. This is a mixture-of-experts model and no card carries active-parameter counts, so these predictions use total parameters and understate the real speed.
DeviceMemoryBandwidthBest qualityWeightstok/sFastest
Cerebras WSE-344.0 GB21000000.0 GB/sbf1615.79 GB~931158.5~3724634.1 q4
NVIDIA GB200 Grace Blackwell Superchip372.0 GB16000.0 GB/sbf1615.79 GB~709.5~2837.8 q4
AMD Instinct MI355X288.0 GB8000.0 GB/sbf1615.79 GB~354.7~1418.9 q4
NVIDIA B200180.0 GB7700.0 GB/sbf1615.79 GB~341.4~1365.7 q4
Google TPU7x (Ironwood)192.0 GB7380.0 GB/sbf1615.79 GB~327.2~1308.9 q4
AMD Instinct MI325X256.0 GB6000.0 GB/sbf1615.79 GB~266.0~1064.2 q4
AMD Instinct MI300X192.0 GB5300.0 GB/sbf1615.79 GB~235.0~940.0 q4
NVIDIA H200 SXM141.0 GB4800.0 GB/sbf1615.79 GB~212.8~851.3 q4
NVIDIA H100 NVL94.0 GB3900.0 GB/sbf1615.79 GB~172.9~691.7 q4
NVIDIA H100 SXM80.0 GB3350.0 GB/sbf1615.79 GB~148.5~594.2 q4
AMD Instinct MI250X128.0 GB3200.0 GB/sbf1615.79 GB~141.9~567.6 q4
Google TPU v5p95.0 GB2765.0 GB/sbf1615.79 GB~122.6~490.4 q4
NVIDIA A100 80GB SXM80.0 GB2039.0 GB/sbf1615.79 GB~90.4~361.6 q4
NVIDIA H100 PCIe80.0 GB2000.0 GB/sbf1615.79 GB~88.7~354.7 q4
NVIDIA GeForce RTX 509032.0 GB1792.0 GB/sbf1615.79 GB~79.5~317.8 q4
Google TPU v6e (Trillium)32.0 GB1638.0 GB/sbf1615.79 GB~72.6~290.5 q4
AMD Instinct MI21064.0 GB1600.0 GB/sbf1615.79 GB~70.9~283.8 q4
NVIDIA A100 40GB SXM40.0 GB1555.0 GB/sbf1615.79 GB~69.0~275.8 q4
Google TPU v432.0 GB1200.0 GB/sbf1615.79 GB~53.2~212.8 q4
NVIDIA GeForce RTX 409024.0 GB1008.0 GB/sbf1615.79 GB~44.7~178.8 q4
AMD Radeon RX 7900 XTX24.0 GB960.0 GB/sbf1615.79 GB~42.6~170.3 q4
NVIDIA GeForce RTX 508016.0 GB960.0 GB/sfp87.89 GB~85.1~170.3 q4
NVIDIA GeForce RTX 309024.0 GB936.0 GB/sbf1615.79 GB~41.5~166.0 q4
NVIDIA L40S48.0 GB864.0 GB/sbf1615.79 GB~38.3~153.2 q4
Apple M3 Ultra512.0 GB819.0 GB/sbf1615.79 GB~36.3~145.3 q4
AMD Radeon RX 7900 XT20.0 GB800.0 GB/sfp87.89 GB~70.9~141.9 q4
Apple M2 Ultra192.0 GB800.0 GB/sbf1615.79 GB~35.5~141.9 q4
Google TPU v5e16.0 GB800.0 GB/sfp87.89 GB~70.9~141.9 q4
NVIDIA GeForce RTX 4080 SUPER16.0 GB736.0 GB/sfp87.89 GB~65.3~130.5 q4
NVIDIA GeForce RTX 4070 Ti SUPER16.0 GB672.0 GB/sfp87.89 GB~59.6~119.2 q4
AMD Radeon RX 9070 XT16.0 GB640.0 GB/sfp87.89 GB~56.8~113.5 q4
Apple M4 Max128.0 GB546.0 GB/sbf1615.79 GB~24.2~96.8 q4
Apple M1 Max64.0 GB400.0 GB/sbf1615.79 GB~17.7~70.9 q4
Apple M2 Max96.0 GB400.0 GB/sbf1615.79 GB~17.7~70.9 q4
Apple M3 Max128.0 GB400.0 GB/sbf1615.79 GB~17.7~70.9 q4
NVIDIA GeForce RTX 3060 12GB12.0 GB360.0 GB/sfp87.89 GB~31.9~63.9 q4
NVIDIA L424.0 GB300.0 GB/sbf1615.79 GB~13.3~53.2 q4
NVIDIA GeForce RTX 4060 Ti 16GB16.0 GB288.0 GB/sfp87.89 GB~25.5~51.1 q4
Apple M4 Pro64.0 GB273.0 GB/sbf1615.79 GB~12.1~48.4 q4
Apple M432.0 GB120.0 GB/sbf1615.79 GB~5.3~21.3 q4

Where it runs

Taken from the card's availability section.

PlatformId on that platform
Hugging FaceinclusionAI/Ling-3.0-tiny

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_lcrLing 3.0 Tiny60.33%2026-09-10 evaluatedindependent evaluatorsource
critptLing 3.0 Tiny0.0%2026-09-10 evaluatedindependent evaluatorsource
gdpval_aaLing 3.0 Tiny10.78%2026-09-10 evaluatedindependent evaluatorsource
gpqa_diamondLing 3.0 Tiny73.43%2026-09-10 evaluatedindependent evaluatorsource
scicodeLing 3.0 Tiny24.19%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