GPT-OSS 120B
CANONICAL MODEL

GPT-OSS 120B

OpenAI · text · released 2025-08-05

Canonical ID
openai/gpt-oss-120b
Open weights
Yes
Status
ACTIVE

Provider routes

GPT-OSS 120B
via Vercel AI Gateway
Input $0.10/MOutput $0.50/M
Tools: YesContext: 131,072Open weights: YesReasoning

Last verified: 2026-09-29 · sources & route detail

GPT-OSS 120B
via Nebius Token Factory
Input $0.15/MOutput $0.60/M
Tools: YesContext: 131,072Open weights: YesReasoning

Last verified: 2026-09-29 · sources & route detail

GPT-OSS 120B
via OCI Generative AI
Input $0.15/MOutput $0.60/M
Tools: YesContext: 128,000Open weights: YesReasoning

Last verified: 2026-09-29 · sources & route detail

GPT-OSS 120B
via Cloudflare Workers AI
Input $0.35/MOutput $0.75/M
Tools: YesContext: 128,000Open weights: YesReasoning

Last verified: 2026-09-29 · sources & route detail

GPT-OSS 120B
via CloudFerro Sherlock
Input $2.92/MOutput $2.92/M
Tools: YesContext: 131,000Open weights: YesReasoning

Last verified: 2026-09-29 · sources & route detail

Use cases

Agentic Coding · Recommended
Route exposes tool calling and is suitable for agent workflows.

Cheapest · Recommended
The recorded provider-route price is low relative to this reference set.

Coding · Recommended
Route exposes tool calling and is suitable for agent workflows.

General · Recommended
Route exposes tool calling and is suitable for agent workflows.

Local/Self Hosted · Recommended
Open weights are documented for this canonical model.

Open Weight · Recommended
Open weights are documented for this canonical model.

Reasoning · Recommended
Route exposes tool calling and is suitable for agent workflows.

Tool Calling · Recommended
Route exposes tool calling and is suitable for agent workflows.

Harness compatibility

Aider · Compatible With Configuration
Provider support: Openrouter. Tool calling: Yes.

Hermes Agent · Compatible With Configuration
Provider support: Openrouter. Tool calling: Yes.

OpenCode · Compatible With Configuration
Provider support: Openrouter. Tool calling: Yes.

Benchmark results

BenchmarkScoreMetricConfidence
LiveBench 2025-0263.4reported scoreLOW

Sources