OpenAI · GPT OSS
gpt-oss-120b
An open-weight mixture-of-experts reasoning model designed for agentic and general-purpose workloads.
Model Markets treats this immutable version as the base entity. Provider endpoints, pricing, quantization, regions, service tiers, and observations remain separate so the record can change without rewriting the model itself.
Provider comparison
Available endpoints
| Provider | Endpoint | Quant. | Input / 1M | Output / 1M | Region | Evidence |
|---|---|---|---|---|---|---|
| CoreWeaveAvailable | coreweave/fp4 | fp4 | $0.030 | $0.17 | Provider global routing | Provider-reported |
| DeepInfraAvailable | deepinfra/bf16 | bf16 | $0.037 | $0.17 | Provider global routing | Provider-reported |
| NovitaAvailable | novita/fp4 | fp4 | $0.050 | $0.25 | Provider global routing | Provider-reported |
| DigitalOceanAvailable | digitalocean | unknown | $0.055 | $0.385 | Provider global routing | Provider-reported |
| Google Vertex AIAvailable | google-vertex/global | unknown | $0.090 | $0.36 | Provider global routing | Provider-reported |
Model specification
Capability record
Capabilities below come from the cited provider catalog. Independent correctness probes are intentionally separate.
Questions
gpt-oss-120b questions
How much does gpt-oss-120b cost?+
The lowest provider-reported input price in this snapshot is $0.030 per million tokens. Output prices begin at $0.17 per million tokens. Compare provider pricing →
Which providers offer gpt-oss-120b?+
5 provider endpoints appear in this source snapshot: CoreWeave, DeepInfra, Novita, DigitalOcean, Google Vertex AI. Browse providers →
Which provider is cheapest for gpt-oss-120b?+
CoreWeave has the lowest listed input price in this snapshot. This is a provider-reported price, not an execution quote.
Which provider is fastest for gpt-oss-120b?+
Model Markets does not yet have approved independent probe results for this model, so it does not name a fastest provider.
What is the gpt-oss-120b context window?+
The cited model record reports a 131.1K-token context window.
Does gpt-oss-120b support tools and structured output?+
The cited provider catalog reports reasoning, tools, structured outputs. These are provider-reported capabilities until independently tested.
Is gpt-oss-120b open-weight?+
gpt-oss-120b is recorded as open-weight under the Apache 2.0 license.
How current is this data?+
The source snapshot was observed Aug 18, 2026. The page is explicitly marked non-live until production collectors are activated.
How does Model Markets measure latency?+
Latency uses disclosed-region active probes, preserving model version, provider, region, workload, service tier, request parameters, and timestamp. No paid probes were run for this delivery. Read the methodology →