Groq & HUMAIN deploy Open AI models on Day Zero

OpenAI's long awaited open source models now live on Groq Cloud

#SaudiArabia #inference - US artificial intelligence infrastructure specialist Groq and Saudi Arabia's new national AI company HUMAIN have launched OpenAI's two new open-weight models gpt-oss-120B and gpt-oss-20B on GroqCloud with day-zero availability worldwide. The collaboration delivers real-time performance with full 128K context length, integrated server-side tools, and local support in Saudi Arabia. The two open source models were released by OpenAI on Tuesday, marking its first open language models since GPT-2's release over five years ago. GroqCloud is offering gpt-oss-120B with pricing from $0.15 per million input tokens and gpt-oss-20B from $0.10 per million input tokens.

SO WHAT? - OpenAI's first open-weight reasoning models in over five years are immediately accessible to Middle East developers and enterprise customers via Groq's regional data centres in Saudi Arabia. The state-of-the-art models offer o-series capabilities with flexible deployment options from consumer laptops to enterprise GPUs. GroqCloud provides Saudi developers and users with day-zero access to the two cutting-edge models without the delays often associated with regional availability.

Here are some key points regarding the models on GroqCloud:

• AI infrastructure innovator Groq and Saudi Arabia’s Public Investment Fund-owned HUMAIN have launched OpenAI's two new open-weight models gpt-oss-120B and gpt-oss-20B on GroqCloud with day-zero availability in the Middle East region and worldwide.

• The launch delivers gpt-oss-120B and gpt-oss-20B with full 128K context, real-time responses, and integrated server-side tools live on Groq’s optimized inference platform GroqCloud, powered by its Dammam data centre.

• The new models are OpenAI's first open-weight language models since GPT-2 over five years ago, with the larger gpt-oss-120B running on single Nvidia GPUs and lightweight gpt-oss-20B operating on consumer laptops with 16GB memory.

• Groq delivers performance speeds of 500+ tokens per second for gpt-oss-120B and 1000+ tokens per second for gpt-oss-20B, with competitive pricing at $0.15/$0.75 per million input/output tokens for the larger model, and $0.10/$0.50 for the 20 billion version.

• Saudi Arabia-based HUMAIN provides local support and access through its position as a PIF company specialising in AI services, strengthening regional developer capabilities and infrastructure access across the Kingdom.

• GroqCloud delivers superfast inference speed, scalability, and cost-efficiency for HUMAIN AI inference customers in Saudi Arabia.

• Groq's global data centre footprint across North America, Europe and the Middle East ensures minimal latency worldwide, with over 1.9 million developers already using the platform for AI inference.

• The new OpenAI models perform well across several open model benchmarks, offering reasoning capabilities similar to its o-series whilst maintaining free availability through Hugging Face.

ZOOM OUT - OpenAI's cutting edge gpt-oss models are built using a mixture-of-experts architecture with the larger model activating just 5.1B parameters per token despite containing 117B total parameters. Trained using reinforcement learning techniques from OpenAI's most advanced internal systems including o3, these models achieve near-parity with proprietary counterparts on reasoning benchmarks whilst offering flexible deployment from edge devices to enterprise infrastructure under the permissive Apache 2.0 licence, potentially reshaping competitive dynamics in the open-source AI landscape.

[Written and edited with the assistance of AI]

LINKS

• GroqCloud (Groq)

• HUMAIN-Groq showcase page (HUMAIN)

• gpt-oss model page (OpenAI)

• gpt-oss code (GitHub)

• gpt-oss 120B (Hugging Face)

• gpt-oss 20B (Hugging Face)

Read more about Groq:

• Llama 4 now live via Middle East GroqCloud service (Middle East AI News)

• Groq opens EMEA's largest AI compute centre in KSA (Middle East AI News)

• Groq ships to Saudi to for huge AI infrastructure cluster (Middle East AI News)

• Groq to offer on-demand services from Saudi Arabia (Middle East AI News)