What GPU do I need to run Aleph-Alpha/Kolibri-1?
78.1B parameters, published in F8_E4M3. View on Hugging Face
Kolibri-1 is published by Aleph-Alpha on Hugging Face, with 4,138 downloads and 689 likes to date. It's a Kolibri1ForCausalLM model built for text-generation, published natively in F8_E4M3.
VRAM required & cheapest live GPU fit
Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.
| Precision | Weight size | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| FP8 (native) | 72.7 GB | 87.3 GB | RTX PRO 6000 | 1 | $1.38/hr |
| cheaper alt. | RTX 5060 Ti | 6 | $1.08/hr | ||
| INT4 (quantized) | 36.4 GB | 43.6 GB | RTX A6000 | 1 | $0.363/hr |
| cheaper alt. | RTX A5000 | 2 | $0.352/hr |
A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.
INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.
Cheapest way to run Kolibri-1 at its published (F8_E4M3) precision: 1× RTX PRO 6000, at $1.38/hr per GPU ($1.38/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Kolibri-1: KV cache by context length
The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.
Sliding-window attention: windowed layers stop growing at the window, the full-attention layers keep growing. Cached per token: 102,400 bytes at 16-bit while the context is inside the window.
| Context | KV cache, one sequence |
|---|---|
| 4K | 0.12 GB |
| 32K | 0.66 GB |
| 128K | 2.5 GB |
| 256K (model maximum) | 5.0 GB |
Computed from the layer, head and window counts in the model's own configuration (Aleph-Alpha/Kolibri-1), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.
Kolibri-1: common questions
Can Kolibri-1 run on a single GPU?
Yes, but not on a desktop card. At FP8 (native) it needs 87.3 GB of VRAM, more than a single 32 GB desktop card holds. The cheapest live single-card fit is a 95.0 GB RTX PRO 6000 at $1.38/hr.
Is Kolibri-1 already quantized?
Yes. It is published in FP8, one byte per parameter, so the 87.3 GB figure is already a quantized footprint rather than a full-precision one. Only INT4 goes below it, at 43.6 GB. A GPU without FP8 tensor cores cannot run it as published, which is why cards here are matched on precision support and not on VRAM alone.
Does quantizing Kolibri-1 lower the GPU bill?
Yes. At FP8 (native) the cheapest live fit is one RTX PRO 6000 at $1.38/hr. At INT4 (quantized) it drops to one RTX A6000 at $0.363/hr, provided a quantized checkpoint exists for it.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More Aleph-Alpha models
- Kolibri-1-BF16 (78.1B, BF16)
Lineage
Based on Kolibri-1-BF16.
Alternatives at this size
Other models for text-generation within about a third of Kolibri-1's 78.1B parameters, from other model lines.
- Qwen-72B (72.3B, BF16)
- Llama-3.3-70B-Instruct (70.6B, BF16)
- GLM-4.5-Air (110.5B, BF16)
- Hunyuan-A13B-Instruct (80.4B, BF16)
- Apertus-70B-Instruct-2509 (70.6B, BF16)
More on Kolibri-1
Fits on a 96 GB GPU at FP8: every model that fits in 96 GB.
Fits on a 48 GB GPU at INT4: every model that fits in 48 GB.
Best chat and assistants models: how Kolibri-1 ranks against the rest.
Related reading: RTX PRO 6000 pricing and specs, How much VRAM you need for LLMs, Serving LLMs with vLLM, vLLM vs TensorRT-LLM vs SGLang, and Best GPU for LLM inference.
Your own GPU cloud accounts
One console for the GPU clouds you already rent from. Your keys stay on your machine.
$20 a month, flat