llamaperf

BitNet b1.58

3 reports

BitNet b1.58 VRAM requirements by size and quant →

How does BitNet b1.58 run on your hardware?

Pick your GPU or Mac, then read the speeds people reported for BitNet b1.58 on it, or estimate memory fit and speed.

Free to use. No account needed. Memory estimates and community measurements are labelled separately.

Run BitNet b1.58 yourself? Add your numbers.

Your coding agent can run a speed test on your machine and send it here with a free account. Or paste a result you've already posted.

Reported performance for BitNet b1.58

Filter this model’s reports by setup →

Speeds are as each source reported them. GPU count, offloading and concurrent requests change them, and prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing two reports.

Tone: mixedAMD hardware
generation:
13.7 tokens/s
quant:
ternary (AIM-W)

User reports BitNet b1.58 2B at 13.68 tok/s on a Radeon Pro W7900 with a self-built C inference stack, AimStack. The GPU backend loads the same 0.87GB ternary file, dequantizes weights on load and runs matmul in fp32; 256 of 256 greedy tokens matched the C engine bit-for-bit. The C engine runs the full model on a 2018 dual-core 8GB laptop at 3.20 tok/s with two workers, and its SIMD path measured 85.53 tok/s. BF16 is 2.4x faster at 32.96 tok/s but flips argmax on near-tied logits, so only FP32 passes validation. The GPU path gets the 78% memory saving but not yet a compute advantage; the UI server drives only the CPU C engine.

Nov 18, 2026

BitNet b1.58 2B

Unknown GPU

generation:
3.2 tokens/s
quant:
1.58-bit ternary

User reports Microsoft's BitNet b1.58-2B running at 3.2 tok/s with 2 workers (2.0 tok/s serial) on a 2018 Dell laptop with an i7-7500U, 8GB RAM and no GPU. The stack is a from-scratch C inference engine with a 1.58-bit ternary core, weights converted to a custom 2-bit packed .aimw format at 0.87 GB. The ternary core itself measured 85.5 tok/s. The run is CPU-only; the laptop's 940MX is deliberately unused. The user notes their own trained models are not released yet.

Nov 16, 2026

BitNet b1.58 2B

Unknown GPU · Project Zero

Tone: positive
generation:
36.0 tokens/s
quant:
ternary (GGUF)

User reports BitNet b1.58-2B-4T at 36 t/s on a Xeon with Project Zero, a self-written C99 inference engine, about 1.8x faster than bitnet.cpp. The engine also runs SmolLM2 F16 at ~100 t/s on an i5-11300H, where llama.cpp is about 7% faster, and DeepSeek Q4_K at 1.9 t/s versus 13.7 t/s for llama.cpp. It is built with GCC and make, uses AVX-512 kernels for ternary packing, and exposes an OpenAI-compatible API with SSE streaming.

Oct 8, 2026

Get a weekly email of new BitNet b1.58 reports on any GPU.

Email me new reports

Reported configurations

Individual observations from this page, not expected speeds or a ranking. Select a model to open its report, with the source and offloading conditions. Reported context may be a limit rather than actual input length.

Model / hardwareQuant / engineReported contextGeneration
BitNet b1.58 2B
AMD Radeon Pro W7900
ternary
AimStack
Not reported13.7 tokens/s