NVIDIA RTX 3060 12GB + Qwen3.8-Flash-Next
IQ1_SStrata
70tok/s
L1 Reproduced12GB VRAM · Ubuntu 24.04.4 LTS
Real data · Developer community · Always updated
What AI can your machine run?
Query the model, framework, and quantization combos your hardware can run — based on real benchmark runs.
Beyond speed: how reliable each data point is, whether it's directly comparable, and where the evidence comes from.
I have hardwareSee which models your machine can runI want to run modelsSee what hardware this model needs
Real runs
From the community and independent tests
Trusted grading
L0 / L1 / L2 evidence tiers
Decision support
Not just scores — actionable advice
3,365
Records
67
Hardware
51
Models
9
Frameworks
3,301
Verified combos
Latest updates
- HardwareNew dual-3090 A/B report: faster long-prompt processing, tighter host RAM headroom
- Runtimellama.cpp v0.9.0 ships FP8 types and expanded audio interfaces, not universal speedups
- RuntimeCommunity measurement tooling fixes locale parsing: why 10.02 GB became 8,00 GB
- Runtimellama.cpp b12079 adds a cross-node NCCL path for tensor parallelism
- HardwareNew dual-3090 report corrects host RAM guidance to 94 GB: GiB and GB are not interchangeable
Popular benchmarks
The hardware × model combos the community cares about
Popular hardware
The hardware devices people watch most
Popular models
Open models the community follows
