# Cheapest gpt-oss-120b API providers, compared (prices checked Nov 7, 2026)

URL: https://ernd.ai/inference-prices/gpt-oss-120b

> ernd partners with [Wafer](https://ernd.ai/partners/wafer). Prices are from OpenRouter's public listings.

gpt-oss-120b is an open-weight model from OpenAI. On OpenRouter, 26 listings from 23 providers sell it, and the blended price runs from $0.025 to $0.45 per million tokens.

| Provider | Input /1M | Output /1M | Blended 3:1 | Cache read /1M | Context | Quantization | Latency | Throughput |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Prime Intellect | $0.015 | $0.055 | $0.025 | $0.013 | 131K | FP4 | 0.84 s | 73 tok/s |
| NEAR AI | $0.03 | $0.15 | $0.06 | $0.015 | 131K | FP4 | 0.64 s | 49 tok/s |
| Venice | $0.03 | $0.15 | $0.06 | $0.03 | 128K | INT4 | 0.83 s | 40 tok/s |
| CoreWeave | $0.03 | $0.17 | $0.065 | $0.03 | 131K | FP4 | 0.27 s | 91 tok/s |
| DekaLLM | $0.03 | $0.18 | $0.0675 | $0.03 | 131K | BF16 | 0.51 s | 53 tok/s |
| AkashML | $0.03 | $0.187 | $0.0693 | $0.03 | 131K | FP4 | 0.39 s | 103 tok/s |
| DeepInfra (bf16) | $0.037 | $0.17 | $0.0703 | – | 131K | BF16 | 0.43 s | 35 tok/s |
| Crusoe | $0.05 | $0.25 | $0.10 | $0.05 | 131K | BF16 | 0.24 s | 217 tok/s |
| Mancer, down when checked | $0.055 | $0.25 | $0.104 | $0.055 | 131K | FP8 | 0.82 s | 48 tok/s |
| Sail Research | $0.06 | $0.40 | $0.145 | $0.03 | 131K | FP4 | 0.62 s | 129 tok/s |
| DigitalOcean | $0.06 | $0.42 | $0.15 | $0.012 | 128K | Not listed | 0.69 s | 51 tok/s |
| Google Vertex (Global), down when checked | $0.09 | $0.36 | $0.158 | – | 131K | Not listed | 0.51 s | 56 tok/s |
| Baseten | $0.10 | $0.50 | $0.20 | $0.10 | 128K | FP4 | 0.18 s | 156 tok/s |
| Baseten (US) | $0.10 | $0.50 | $0.20 | $0.10 | 128K | FP4 | 0.17 s | 165 tok/s |
| Amazon Bedrock | $0.15 | $0.60 | $0.262 | – | 131K | Not listed | 0.51 s | 210 tok/s |
| Amazon Bedrock (EU-WEST-1) | $0.15 | $0.60 | $0.262 | – | 131K | Not listed | 0.48 s | 152 tok/s |
| Groq, down when checked | $0.15 | $0.60 | $0.262 | $0.075 | 131K | Not listed | 0.19 s | 342 tok/s |
| Nebius Token Factory | $0.15 | $0.60 | $0.262 | $0.10 | 131K | FP4 | 0.27 s | 281 tok/s |
| Phala | $0.15 | $0.60 | $0.262 | – | 131K | Not listed | 0.93 s | 145 tok/s |
| SiliconFlow, down when checked | $0.15 | $0.60 | $0.262 | $0.075 | 131K | FP8 | 1.79 s | 38 tok/s |
| Together | $0.15 | $0.60 | $0.262 | – | 131K | Not listed | 0.34 s | 91 tok/s |
| Parasail | $0.10 | $0.75 | $0.263 | $0.055 | 131K | FP4 | 0.37 s | 140 tok/s |
| Parasail (Fast) | $0.15 | $0.75 | $0.30 | – | 131K | FP4 | 0.28 s | 162 tok/s |
| SambaNova | $0.14 | $0.95 | $0.343 | – | 131K | Not listed | 1.10 s | 335 tok/s |
| DeepInfra (fp8), down when checked | $0.20 | $0.95 | $0.388 | – | 131K | FP8 | 1.42 s | 59 tok/s |
| Cerebras | $0.35 | $0.75 | $0.45 | $0.35 | 131K | FP16 | 0.19 s | 807 tok/s |

Blended is three parts input price to one part output, per million tokens. Latency is time to first token and throughput is output tokens per second, both averaged from OpenRouter's daily figures for the last 7 days. Cheapest and fastest only count listings that were up when checked.

## Questions

### What is the cheapest gpt-oss-120b API provider?

Prime Intellect, at $0.025 per million tokens blended ($0.015 input, $0.055 output), in OpenRouter's listings on November 7, 2026.

### How much does gpt-oss-120b cost per million tokens?

Input runs from $0.015 to $0.35 per million tokens across 26 listings, and output from $0.055 to $0.95. The middle listing comes to $0.20 blended.

### Which gpt-oss-120b provider is fastest?

Cerebras had the most throughput over the last 7 days, about 807 tokens per second. Baseten (US) was quickest to the first token at 0.17 s.

### How do I pay for gpt-oss-120b inference?

Buy OpenRouter credits and it sends each request to one of these providers, or open an account with a provider and pay it directly. The ernd card is a credit card for US companies. It earns 4% back on AI vendor spend and 1.5% back on everything else.

Source: OpenRouter public model and provider listings (https://openrouter.ai/models), checked November 7, 2026.

The ernd card is a credit card for US companies. It earns 4% back on AI vendor spend and 1.5% back on everything else. [Request the card at app.ernd.ai](https://app.ernd.ai/login?next=%2Fapp%2Fcard%2Fapply&utm_source=llms&utm_campaign=inference_prices).