Immediately start using gpt-oss-120B on Together AI's production-ready infrastructure with proven performance, high reliability and comprehensive developer toolingToday, OpenAI has released gpt-oss-120B and gpt-oss-20B under Apache 2.0 licensing. Access these models on Together AI to democratize AI deployment across industries, communities, and countries globally.TL;DR:OpenAI's new open-weight models gpt-oss-120B and gpt-oss-20B, released under Apache 2.0 license for unlimited modification and deployment, on Together AIAvailable as serverless endpoints or dedicated endpoints with instant deploymentProven performance and superior economics with optimized inference on Together's infrastructureHigh reliability at scale with 99.9% uptime SLA and global deploymentComprehensive developer tooling through unlimited fine-tuning and complete AI engineer control{{custom-cta-3}}Together AI's dedicated infrastructure delivers the enterprise reliability that The Washington Post requires for deploying AI applications into production. When breakthrough models like OpenAI's open-weight models become available, Together AI's platform ensures we can deploy them on day 1 with the performance standards our organization needs. — Nick Huber, AI Ecosystem Lead, QuoraOpenAI's gpt-oss-120B and gpt-oss-20B: Purpose-Built, Open-Weight Reasoning ModelsOpenAI has released gpt-oss-120B and gpt-oss-20B as open models designed for reasoning tasks. These aren't just scaled-up versions of existing models — they're purpose-built open-weight reasoning models released under Apache 2.0 licensing while maintaining the safety standards that enterprise deployment requires.The development process reflected OpenAI's commitment to global accessibility. Through listening sessions across the U.S., Europe, and Asia, they gathered feedback to ensure these models lower barriers for emerging markets, resource-constrained sectors, and smaller organizations. Each model underwent rigorous safety training and evaluation, with OpenAI testing maliciously fine-tuned versions under their Preparedness Framework to verify safety even after weight release.Model specifications: 📊 Benchmark 🤖 gpt-oss-120B 🪶 gpt-oss-20B 🏛 OpenAI o3 📈 OpenAI o4-mini Reasoning & knowledge MMLU 90.0 85.3 93.4 93.0 GPQA Diamond 80.1 71.5 83.3 81.4 Humanity’s Last Exam 19.0 17.3 24.9 17.7 Competition math AIME 2024 96.6 96.0 95.2 98.7 AIME 2025 97.9 98.7 98.4 99.5 What makes gpt-oss-120B and gpt-oss-20B transformative: Complete freedom to fine-tune for specialized domains, deploy anywhere, and build commercial applications with full model ownership.Deploy gpt-oss-120B on Together's Proven InfrastructureOpenAI's gpt-oss-120B is available serverlessly on Together AI, removing the friction of deploying large reasoning models. No infrastructure setup. No throttling. Just access via standard APIs but with the performance guarantees and economics that enable global innovation.💰Price-Performance Leadership ✓ Superior economics vs proprietary alternatives ✓ Faster inference through research-driven optimizations ✓ Transparent pricing with no hidden fees 🌍Reliable at Scale ✓ 99.9% uptime SLA with multi-region deployment ✓ Zero throttling during viral traffic spikes ✓ SOC 2 compliance for enterprise deployment 🛠️Built for AI Engineers ✓ Research-grade tooling in a unified platform ✓ Fine-tune, distill and deploy models ✓ OpenAI-compatible APIs for easy migration Together's frontier AI cloud is designed for deploying the world's most advanced AI models with uncompromising performance and high reliability. For this launch, we also partnered closely with NVIDIA – research and engineering teams across both Together AI and NVIDIA worked on unlocking greater performance gains for our users. When you immediately start using gpt-oss-120B with us, you get:Start experimenting immediately: Access gpt-oss-120B through our playground and chat application to test capabilities, then move seamlessly to production APIs when ready.Superior economics that make sense: Deploy gpt-oss-120B at ＄0.15 per 1M input tokens and ＄0.60 per 1M output tokens. Use our Batch API for even more cost-effective distillation and synthetic data generation.Deploy with proven reliability: 99.9% availability SLA with multi-region deployment and enterprise security hosted on SOC 2 compliant servers in North America ensures your agentic workflows complete successfully, even during unexpected traffic surges.Scale without limits: Automatic scaling from serverless to dedicated clusters handles everything from prototyping to full production during peak traffic, without throttling.Comprehensive optimization for your needs: Our research team's innovations — from FlashAttention to custom kernels — deliver up to 50% cost savings and 2x performance improvements. Access gpt-oss-120B as a dedicated endpoint for custom fine-tuning, or use our Fine-Tuning API to tailor the model to your specific use cases.This isn't just model hosting. Together AI offers a comprehensive platform built to make cutting-edge AI models like gpt-oss-120B production-ready from day one with the security, proven performance, and cost advantages that enterprise applications require.We've been consistently impressed by Together's ability to quickly provide us with access to the latest open source models - often on day 1 of release - all while delivering exceptional performance and cost-efficiency." — Nick Huber, AI Ecosystem Lead, QuoraReal-World Applications & Getting StartedOpenAI's gpt-oss-120B and gpt-oss-20B unlock applications across domains:Emerging Market Innovation: Organizations in resource-constrained regions can fine-tune gpt-oss-120B and gpt-oss-20B for local languages and deploy at costs that enable sustainable business models.Government Digital Sovereignty: Deploy gpt-oss-120B and gpt-oss-20B in air-gapped environments with complete data control while maintaining proven production performance.Research and Education: Academic institutions gain full transparency into gpt-oss-120B and gpt-oss-20B architecture and training approaches — enabling research impossible with proprietary models.Enterprise Customization: Adapt gpt-oss-120B and gpt-oss-20B for specialized workflows and integrate with existing systems without vendor dependencies.Start using gpt-oss-120B through Together AI's production APIs: