GPT-OSS 120B API pricing

20 providers serve GPT-OSS 120B. GPT-OSS 120B is an open-weight model (Apache-2.0) with 117B total parameters, 5.1B active per token (mixture-of-experts), up to 131K context. Prices below are per 1M tokens, cheapest input first (self-hosting math further down).

Refreshed 2026-08-23 · list prices from provider APIs
Input floor $0.050/1M at DeepInfra and Novita AI Output floor $0.250/1M at Novita AI Find your workload winner →
Provider Input $/1M Output $/1M Cache read $/1M Context
DeepInfra $0.050 $0.450 - 131K Use
Novita AI $0.050 $0.250output floor - 131K Use
OVHcloud $0.080 $0.400 - 131K Use
Google Vertex AI $0.090 $0.360 - 131K Use
Baseten $0.100 $0.500 - - Use
Azure AI Foundry $0.150 $0.600 - 131K Use
Databricks $0.150 $0.600 - 131K Use
Fireworks AI $0.150 $0.600 $0.015 131K Use
Groq $0.150 $0.600 $0.075 131K Use
Together AI $0.150 $0.600 - 131K Use
IBM watsonx $0.150 $0.600 - 8K Use
Scaleway $0.150 $0.600 - 128K Use
AWS Bedrock $0.150 $0.600 - 131K Use
Tensormesh $0.150 $0.600 - 131K
OpenRouter $0.180 $0.800 - 131K Use
Replicate $0.180 $0.720 - - Use
SambaNova $0.220 $0.590 - 131K Use
Cerebras $0.350 $0.750 - 131K Use
Cloudflare $0.350 $0.750 - 128K Use
Crusoe $0.800 $0.800 - 131K Use

The cheapest way to run GPT-OSS 120B

On an illustrative 70M-input / 30M-output monthly workload, today's lowest listed cost is roughly $11.00/month, at Novita AI. Input-heavy, output-heavy and cache-heavy workloads can produce different winners. Use the live mix calculator below rather than combining floors from two different providers. On output-token cost alone, a busy self-hosted deployment can work out cheaper (~$0.035/1M output on MI300X). See the math below. Run your own numbers in the breakeven calculator, or read the breakeven math.

Your actual monthly cost

Uses today's list prices

Cache read uses the listed cache rate where available; otherwise it falls back to normal input price. Batch, write-cache, volume and negotiated discounts are excluded.

Or run it yourself: GPT-OSS 120B self-hosting economics

GPT-OSS 120B is open-weight (Apache-2.0), so the API price above competes with the GPU-hour market. 117B total parameters (MoE, ~5.1B active per token) needs roughly 161 GB VRAM at FP8 or 81 GB at INT4, KV-cache headroom included.

GPU that fits (FP8, single card)VRAMFrom $/hrCheapest at
MI300X 192 GB $0.500community RunPod Rent
B200 180 GB $3.06spot Verda
B300 288 GB $3.75spot Verda
GB300 288 GB $4.31spot Verda
Breakeven estimate: a well-batched vLLM deployment on one MI300X ($0.500/hr) at ~7942 aggregate tok/s and 50% utilization works out to roughly $0.035 per 1M output tokens, versus $0.250 via the cheapest output-token API (Novita AI). Self-hosting wins on cost if you can keep the GPU busy. Planning estimate: throughput varies with hardware, quantization, batch size and context. Tune it in the calculator.

Related models

ModelCheapest in $/1MCheapest out $/1MProviders
GPT-OSS 20B $0.015 $0.070 14
GPT-5 $1.25 $10.00 8
GPT-5 Mini $0.250 $1.20 7
GPT-5 Nano $0.050 $0.400 7
GPT-4o $2.50 $10.00 6
GPT-4o-Mini $0.150 $0.600 6

FAQ

What is the cheapest API for GPT-OSS 120B?

As of 2026-08-23, the lowest input price for GPT-OSS 120B, $0.050 per 1M tokens, is the same list price at DeepInfra and Novita AI. The lowest output price is Novita AI at $0.250 per 1M output tokens. The cheapest provider for a real workload depends on its input/output mix. The next-lowest input price is OVHcloud at $0.080, a 60% difference.

How much VRAM do you need to self-host GPT-OSS 120B?

GPT-OSS 120B has 117B parameters, so plan for roughly 161 GB of VRAM at FP8 or 81 GB at INT4/AWQ, KV-cache headroom included. A single MI300X (192 GB) fits it.

← All models · GPU rental prices · Breakeven calculator · Get this data as JSON