GPT-OSS-120B listed rates. Same model name does not prove equal serving quality. What the price covers →
8 more providers · other models and ways to buy
OA
OpenAI
GPT-6.1 Sol Standard$2 input / $10 output per 1M tokens gpt-6.1-sol API text tokens, Standard. Cached input $0.10; cache write $2.50. Over 272K input tokens: full request input/cache rates 2×, output 1.5×. Regional processing adds 10% where offered.Source ↗
Tool callingRequires Responses API for GPT-6.1 Sol gpt-6.1-sol. Context 1,050,000 and max output 128,000 tokens. Do not infer Chat Completions tool parity from an OpenAI-compatible request shape.Source ↗
AN
Anthropic
Sonnet 5.5 base tokens$2 input / $10 output per MTok Claude Sonnet 5.5 first-party standard API rates. 5-minute cache write $2.50, 1-hour $4, cache hit $0.10 per MTok. Cache writes and hits have separate rates. Different model from the GPT-OSS cost illustration.Source ↗
Fable / Mythos 5.230-day retention required unless expressly authorized Claude Fable 5.2 and Claude Mythos 5.2. Not available under ZDR unless Anthropic expressly authorizes it. Do not generalize to every Claude model.Source ↗
GV
Google Vertex AI
Flash introductory tokens$0.75 input / $3.75 output per 1M tokens Gemini 3.8, 3.7 and 3.6 Flash; CodeMender using these models. USD introductory rates through December 31, 2026. Announced January 1, 2027 standard rates $1.50 input / $7.50 output. Separate model from the GPT-OSS illustration.Source ↗
Failed long completionsGemini 4 retrievable usage can still be charged Gemini 4 completion above 32K tokens. Effective November 13, 2026: retrievable token usage is charged even if the request ultimately fails due to client/server timeout. This is an exception to the page’s general non-200 statement.Source ↗
AB
Amazon Bedrock
Retention modenone blocks a model that requires retention Account per Region; project setting only on bedrock-mantle. New scope inherits model default. store=false alone is not ZDR. none prohibits durable request/response storage; Responses store=true rejected and background mode unavailable.Source ↗
Fable 5 / 5.1 / 5.2 on BedrockAWS may retain and review content for up to 30 days Models requiring aws_review, currently Claude Fable 5, 5.1, 5.2. Retained within the AWS boundary; not shared with model provider. Cross-region inference stores retained inputs/outputs in destination processing Regions. ZDR exceptions require per-account/per-model eligibility.Source ↗
MF
Microsoft Foundry
Processing scopeGlobal / Data Zone / Azure geography Foundry Serverless API deployment types, not managed compute. At-rest data stays in designated Azure geography. Global can process any Azure region; Data Zone within US/EU/APAC zone; Standard/Regional Provisioned within geography, possibly between its regions.Source ↗
Global Batch50% discount; 24-hour processing Global Batch, selected supported models. Processing may occur in any Azure region. Data Zone Batch stays within its zone. Not every model supports every deployment type.Source ↗
OR
OpenRouter
Platform feeStandard 5.5% / Business 8% OpenRouter paid plans, underlying model token rates separate. Model tokens billed at posted per-model input/output rates. BYOK has separate fee thresholds. This is not a per-token rate or application total.Source ↗
In-region routingBusiness and Enterprise; US/EU OpenRouter US/EU in-region routing. Standard has data-policy routing but not these in-region features. Contractual SLA Enterprise only; paid rate limits pass through underlying provider.Source ↗
RE
Replicate
Public vs private servingPrivate models can bill setup and idle time Replicate private models except fast-booting fine-tunes. Public models bill processing by time/hardware or input/output. Private models bill setup+idle+active while online; fast-booting fine-tunes are an explicit active-only exception.Source ↗
CE
Cerebras
GPT-OSS-120B PayGo limits500K uncached TPM / 1K RPM Cerebras Developer PayGo, GPT-OSS-120B. Models and limits may change without notice. Custom higher limits, SLAs and dedicated capacity require agreement; no production suitability inferred from a self-serve limit.Source ↗
Prepaid creditsPurchased credits expire after one year Cerebras PayGo purchased credits. At zero balance rate limits revert to zero until funds added. The separate promotional-credit terms are in IA-040.Source ↗
The output changes the bill.
Try your input and output counts against the listed GPT-OSS-120B rates.