Model library / cosmicoptima
Computer-10 is a 70-billion-parameter Llama 3.1-based document model that generates conversational turns without instruction tuning.
Parameters in total
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
Computer-10 is a 70.6 billion-parameter Llama 3.1 derivative trained on a curated anthology of documents. The publisher describes it as a document-oriented model that produces its own personality and opinions, rather than a servile chat assistant, and it expects users to provide a formatted conversation document as input.
In practice the model requires substantial resources: the weights are stored in float32 format and occupy roughly 141 GB, and the model supports a context window of up to 131 072 tokens. It is released under the llama3.1 licence, and deployment must accommodate the large parameter count and extensive context length.
Specification
| Repository | cosmicoptima/computer-10 |
| Publisher | cosmicoptima |
| Published | 22 September 2026 |
| Architecture | LlamaForCausalLM |
| Model type | llama |
| Parameters | 70,553,706,496 |
| Layers | 80 |
| Hidden size | 8,192 |
| Attention heads | 64, 8 key/value heads |
| Context length | 131,072 tokens |
| Vocabulary | 128,256 tokens |
| Weight format | float32 |
| Licence | llama3.1 |
| Base model | cosmicoptima/computer-9 |
Memory
Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.
| Precision | Weights | Notes |
|---|---|---|
| bf16 | 131.4 GiB | Full precision as released |
| fp8 | 65.7 GiB | 8-bit, near-lossless on most models |
| int4 | 32.9 GiB | 4-bit, smallest footprint |
The cache costs 320 KiB per token at 16-bit, so the full 131,072-token context of one request needs about 40.0 GiB. Concurrency multiplies that number, not the weights.
Hardware
At the released precision, with 90% of the card given to the server and 2.5 GiB kept for the runtime. The context column is what is left for the cache on that setup.
| GPU | Cards needed | Free for cache | Context it holds |
|---|---|---|---|
| RTX 4090 24GB | 8 | 21.4 GiB | about 70,069 tokens |
| RTX 5090 32GB | 8 | 79.0 GiB | the full 131,072 tokens |
| L40S 48GB | 4 | 31.4 GiB | about 102,837 tokens |
| A100 80GB | 2 | 7.6 GiB | about 24,849 tokens |
| H100 80GB | 2 | 7.6 GiB | about 24,849 tokens |
| RTX PRO 6000 96GB | 2 | 36.4 GiB | about 119,221 tokens |
| H200 141GB | 2 | 117.4 GiB | the full 131,072 tokens |
Serving
A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.
vllm serve cosmicoptima/computer-10 \ --tensor-parallel-size 8 \ --max-model-len 69632 \ --gpu-memory-utilization 0.90
Questions
The weights take about 131 GiB at the released precision. No single card in the table above holds that, so it needs several GPUs or a lower precision.
The repository declares llama3.1. Read the licence text before commercial use: the name alone does not tell you what is allowed.
The configuration allows 131,072 tokens. Whether the whole window is usable depends on the memory left for the cache, which the hardware table works out per card.
This page is a reference, and Aiqre does not serve this model today. We deploy open-weight models on dedicated EU hardware on request, so if you want an OpenAI-compatible endpoint for it, email [email protected] with the model name and roughly what volume you expect.
Also in the library
Kolibri is a 78.1 billion-parameter mixture-of-experts language model from Aleph-Alpha, supporting German and English with long-c…
AliceAI-Foundation-80B-A3B-Base is an 81-billion-parameter autoregressive language model with mixture-of-experts layers, trained …
Ling-3.0-flash-Fin is a 127.5 billion-parameter finance-enhanced language model for long-context financial research and analysis.
K2-Horizon-MoVA-36B-A4B is a sparse Mixture-of-Experts language model with MoVA attention, released by IFM.
Hosting
Tell us the model and roughly what volume you expect. We reply with a price and an OpenAI-compatible endpoint, with your prompts never stored.