Gpt Oss 120b API - Demo - DeepInfra (https://deepinfra.com/openai/gpt-oss-120b) citeturn144view1 [wordlim: 200] Crawled: today; Content type: text/html; Source: open({"ref_id":"https://deepinfra.com/openai/gpt-oss-120b","lineno":null}); Total lines: 180 L0: DeepInfra raises $107M Series B to scale the inference cloud — 【0†read the announcement】 L1: L2: 【1†Models】 L3: L4: By Category L5: L6: * * * L7: L8: 【2†Automatic Speech Recognition】【3†Other】【4†Decision】【5†Embeddings】【6†Reranker】【7†Text Generation】【8†Text To Image】【9†Text To Music】【10†Text To Speech】【11†Text To Video】【12†World Model】【13†Zero Shot Image Classification】 L9: L10: 【1†View all models】 L11: L12: By Family L13: L14: * * * L15: L16: 【14†/Claude】【15†/DeepSeek】【16†/Flux】【17†/Gemini】【18†/GLM】【19†/Kimi】【20†/Llama】【21†/Mistral】【22†/Nemotron】【23†/Qwen】 L17: 【24†Docs†docs.deepinfra.com】【25†Pricing】【26†GPUs】【27†Chat】【28†DeepStart】【29†DeepCluster】【30†Blog】 L18: L19: 【31†Contact sales】【32†Log in】 L20: L21: 【1†Models】 L22: L23: 【2†Automatic Speech Recognition】【3†Other】【4†Decision】【5†Embeddings】【6†Reranker】【7†Text Generation】【8†Text To Image】【9†Text To Music】【10†Text To Speech】【11†Text To Video】【12†World Model】【13†Zero Shot Image Classification】 L24: L25: 【24†Docs†docs.deepinfra.com】 L26: L27: 【25†Pricing】 L28: L29: 【26†GPUs】 L30: L31: 【27†Chat】 L32: L33: 【28†DeepStart】 L34: L35: 【29†DeepCluster】 L36: L37: 【30†Blog】 L38: L39: Feedback L40: L41: 【31†Contact sales】 L42: 【32†Log in】 L43: L44: 1. 2. 3. 【1†Models】 L45: 4. 5. 【33†Text Generation】 L46: L47: 【34†Image: openai logo】 L48: L49: openai/ L50: # gpt-oss-120b L51: L52: $0.037 L53: L54: in L55: L56: $0.17 L57: L58: out L59: L60: / 1M tokens L61: L62: gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. The model supports configurable reasoning depth, full chain-of-thought access, and native tool use, including function calling, browsing, and structured output generation. L63: L64: Public L65: L66: Zero retention L67: L68: bfloat16 L69: L70: 131,072 L71: L72: JSON L73: L74: Function L75: 【35†Project†huggingface.co】【36†License†huggingface.co】 L76: L77: 【37†Image: openai/gpt-oss-120b cover image†shared.deepinfra.com】 L78: L79: [Button: demo]【38†api】 L80: L81: 【39†Image: openai/gpt-oss-120b cover image】 L82: ###### gpt oss 120b L83: L84: Ask me anything L85: L86: 0.00 s L87: L88: You need to log in to use this model L89: L90: 【32†Log in】 L91: L92: Settings L93: L94: [Button: Reset defaults] L95: L96: Model Information L97: L98: Welcome to the gpt-oss series, 【40†OpenAI’s open-weight models†openai.com】 designed for powerful reasoning, agentic tasks, and versatile developer use cases. L99: L100: We’re releasing two flavors of these open models: L101: * `gpt-oss-120b` — for production, general purpose, high reasoning use cases that fit into a single H100 GPU (117B parameters with 5.1B active parameters) L102: * `gpt-oss-20b` — for lower latency, and local or specialized use cases (21B parameters with 3.6B active parameters) L103: L104: Both models were trained on our 【41†harmony response format†github.com】 and should only be used with the harmony format as it will not work correctly otherwise. L105: > [!NOTE] This model card is dedicated to the larger `gpt-oss-120b` model. Check out 【42†`gpt-oss-20b`†huggingface.co】 for the smaller model. L106: # Highlights L107: * Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. L108: * Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. L109: * Full chain-of-thought: Gain complete access to the model’s reasoning process, facilitating easier debugging and increased trust in outputs. It’s not intended to be shown to end users. L110: * Fine-tunable: Fully customize models to your specific use case through parameter fine-tuning. L111: * Agentic capabilities: Use the models’ native capabilities for function calling, 【43†web browsing†github.com】, 【44†Python code execution†github.com】, and Structured Outputs. L112: * Native MXFP4 quantization: The models are trained with native MXFP4 precision for the MoE layer, making `gpt-oss-120b` run on a single H100 GPU and the `gpt-oss-20b` model run within 16GB of memory. L113: * * * L114: # Reasoning levels L115: L116: You can adjust the reasoning level that suits your task across three levels: L117: L118: * Low: Fast responses for general dialogue. L119: * Medium: Balanced speed and detail. L120: * High: Deep and detailed analysis. L121: L122: The reasoning level can be set in the system prompts, e.g., "Reasoning: high". L123: L124: # Tool use L125: L126: The gpt-oss models are excellent for: L127: L128: * Web browsing (using built-in browsing tools) L129: * Function calling with defined schemas L130: * Agentic operations like browser tasks L131: # Fine-tuning L132: L133: Both gpt-oss models can be fine-tuned for a variety of specialized use cases. L134: L135: This larger model `gpt-oss-120b` can be fine-tuned on a single H100 node, whereas the smaller 【42†`gpt-oss-20b`†huggingface.co】 can even be fine-tuned on consumer hardware. L136: L137: 【45†Image: Footer Logo】 L138: L139: 【46†Image: SOC 2 Certified†static.sprinto.com】【47†Image: ISO 27001 Certified†static.sprinto.com】 L140: L141: Have questions or need a custom solution? L142: L143: 【31†Contact sales】 L144: L145: Company L146: L147: 【25†Pricing】 L148: L149: 【24†Docs†docs.deepinfra.com】 L150: 【48†Compare】 L151: L152: 【28†DeepStart】 L153: L154: 【49†About】 L155: L156: 【50†Careers】 L157: L158: 【31†Contact us】 L159: L160: 【51†Media Center】 L161: L162: 【52†Trust Center†trust.deepinfra.com】 L163: L164: 【53†Status†status.deepinfra.com】 L165: L166: 【54†DeepGPT†deepgpt.com】 L167: L168: Latest Models L169: L170: 【55†thinkingmachines/Inkling-2-Small】【56†nace-ai/drex-max-finance】【57†Lightricks/LTX-Upscale】【58†Lightricks/LTX-Reframe】【59†Lightricks/LTX-HDR】 L171: L172: Featured Models L173: 【60†nvidia/NVIDIA-Nemotron-3-Super-120B-A12B】【61†moonshotai/Kimi-K3】【62†Qwen/Qwen3.6-35B-A3B】【63†Qwen/Qwen3-TTS-VoiceDesign】【64†XiaomiMiMo/MiMo-V2.6-Pro】 L174: L175: 【65†Image: Built With Love in Palo Alto】 L176: L177: © 2026 DeepInfra. All rights reserved. L178: L179: 【66†Privacy Policy】【67†Terms of Service】