General
Issue

OT

Otávio Mariano@tavinckn

New Model / Add model to leaderboard

🔹 Model Name

VICK

🔹 Organization / Author

TechLifes

🔹 Model Type

  • Proprietary

  • Closed / API-only

🔹 Release Date

Not defined

🔹 Model Description

VICK is a large-scale language model developed in Brazil and 100% focused on the Portuguese language, designed to operate at high performance in complex problem solving, advanced conversational reasoning, and code creation/editing.

The model was engineered to handle long-context interactions, maintain logical coherence across extended dialogues, and assist in software development tasks with a strong emphasis on clarity, correctness, and structured reasoning. VICK is optimized for real-world usage scenarios, including analytical workflows, conversational agents, and developer-facing tools.

⚙️ Technical Details

  • Parameters: 106B

  • Context Window: 128k tokens

  • Architecture: Transformer-based (served via vLLM)

  • Training Data: Combination of proprietary datasets and curated public data

  • License: Proprietary

📊 Benchmarks (with sources)

Benchmarks were executed using full, standard evaluation sets.
No reduced question sets or partial evaluations were used.
Results remained consistent across complete benchmark runs.

link: https://benchmark.vick.ia.br/


Benchmark methodology:

  • Evaluation type: Full benchmark runs

  • Prompting strategy: Standardized evaluation prompts

  • Inference setup: Production-equivalent deployment

🌐 Links & Resources

  • Official Website / Model Page: https://vick.ia.br

  • Technical Paper / Report: N/A

  • GitHub / Weights: Not publicly available

  • API Endpoint: Available upon request

🚀 Deployment / Availability

  • Access Method:

    • API

    • Weights download

    • Both

  • Inference Infrastructure:
    Custom infrastructure using vLLM for high-throughput, long-context inference.

📝 Additional Notes

VICK was designed with a strong focus on the Brazilian and Portuguese-speaking ecosystem, aiming to reduce dependency on foreign-language–centric models while maintaining competitive performance against leading international LLMs.

✅ Request

I would like this model to be added to LLM Stats and included in the appropriate leaderboards.

Thank you.

0
0comments

Discussion

0