Recursive Language Model 198M — Hardware Requirements & GPU Compatibility
Chat
Recursive Language Model 198M is a 198M-parameter open language model from Girinath11. It supports a context window of up to 512 tokens. At BF16 it needs about 0.44 GB of VRAM — see which GPUs and Macs can run it below.
787 downloads
10 likesSpecifications
- Publisher
- Girinath11
- Parameters
- 198M
- Architecture
- RecursiveLanguageModel
- Context Length
- 512 tokens
- Vocabulary Size
- 50,260
- Release Date
- 2026-03-13
- License
- MIT
Get Started
HuggingFace
How Much VRAM Does Recursive Language Model 198M Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 0.4 GB | — | 0.40 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Recursive Language Model 198M?
BF16 · 0.4 GBRecursive Language Model 198M (BF16) requires 0.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 1+ GB is recommended. 35 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti.
Runs great
— Plenty of headroomWhich Devices Can Run Recursive Language Model 198M?
BF16 · 0.4 GB33 devices with unified memory can run Recursive Language Model 198M, including NVIDIA DGX H100, NVIDIA DGX A100 640GB.
Runs great
— Plenty of headroomRelated Models
Frequently Asked Questions
- How much VRAM does Recursive Language Model 198M need?
Recursive Language Model 198M requires 0.4 GB of VRAM at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 198M × 16 bits ÷ 8 = 0.4 GB
VRAM usage by quantization
BF160.4 GB- Can I run Recursive Language Model 198M on a Mac?
Recursive Language Model 198M requires at least 0.4 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Recursive Language Model 198M locally?
Yes — Recursive Language Model 198M can run locally on consumer hardware. At BF16 quantization it needs 0.4 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Recursive Language Model 198M?
At BF16, Recursive Language Model 198M can reach ~6625 tok/s on AMD Instinct MI300X. On NVIDIA GeForce RTX 4090: ~1489 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 0.4 × 0.55 = ~6625 tok/s
Estimated speed at BF16 (0.4 GB)
~6625 tok/s~1489 tok/s~4952 tok/s~4096 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Recursive Language Model 198M?
At BF16, the download is about 0.40 GB.
- Which GPUs can run Recursive Language Model 198M?
35 consumer GPUs can run Recursive Language Model 198M at BF16 (0.4 GB). Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 35 GPUs have plenty of headroom for comfortable inference.
- Which devices can run Recursive Language Model 198M?
33 devices with unified memory can run Recursive Language Model 198M at BF16 (0.4 GB), including Mac Mini M4 (16 GB), Mac Mini M4 (32 GB), Mac Mini M4 Pro (24 GB), Mac Mini M4 Pro (48 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.