OpenAI: gpt-oss-120b

openai/gpt-oss-120b

gpt-oss-120b is a 120B open-weight model from OpenAI, ranking #156 of 263 on the TryAii Score. It's strongest on math (#16 on AIME-2025). API pricing starts at $0.15/$0.6 per million input/output tokens — about half of what models at its level usually charge, with 19 independent hosts serving it. It's among the fastest models we track, at roughly 76-247 tokens/sec.

Prompt $/1M

$0.150

Completion $/1M

$0.600

Context

131,072 tok

Best provider

814 tok/s · Cerebras

Specs

Model id: openai/gpt-oss-120b
Display name: OpenAI: gpt-oss-120b
Modality: text->text
Tokenizer: GPT
Instruct type: —
Context length: 131,072
Max completion: 65,536
Prompt $/1M: $0.150
Completion $/1M: $0.600
Free tier: no

Capabilities

Function callingno
JSON modeno
System promptno
Visionno
Moderatedno

Benchmark scores

Each axis is one benchmark, normalized to the best model in the field.

No benchmark data for this model.

Speed by provider

Provider	tok/s	TTFT	Total latency
Cerebrasfastest	814	190 ms	231 ms
SambaNova	383	994 ms	1,322 ms
Groq	311	198 ms	276 ms
Crusoe	303	200 ms	246 ms
Nebius	275	272 ms	365 ms
Artificial Analysis	259	495 ms	—
Mara	235	967 ms	1,137 ms
BaseTen	223	274 ms	318 ms
Novita	204	735 ms	829 ms
Parasail	200	356 ms	426 ms
Together	199	232 ms	303 ms
Phala	168	984 ms	1,124 ms
Google	167	251 ms	640 ms
AkashML	104	464 ms	573 ms
Amazon Bedrock	101	607 ms	654 ms
SiliconFlow	90	1,109 ms	1,280 ms
DekaLLM	77	485 ms	592 ms
DeepInfra	74	223 ms	267 ms
Mancer 2	70	557 ms	635 ms
DigitalOcean	60	348 ms	429 ms
Near AI	49	700 ms	870 ms
CoreWeave	47	400 ms	475 ms
Venice	22	1,979 ms	2,396 ms