OpenAI · in Borg

gpt-oss-120b

gpt-oss-120b is the larger open-weight member of the gpt-oss family.

Reasoning131,072 tokenstext

131,072 tokensContext window

6121 msMedian time to first token (gateway proxy)

46.8Output tokens per second (whole turn)

100%Request success

—Vendor API price in / out · per 1M

Included in Borg

Download Borg Borg for Business

At a glance

Specification	Value
Provider	OpenAI
Family	gpt-oss
Release date	2025-08-05 Vendor source
Context window	131,072 tokens Vendor source
Input modalities	text
Output modalities	text
Reasoning	Yes
Open weights	Yes
Knowledge cutoff	2024-06-01 Vendor source
Borg lane ids	`antigravity/gpt-oss-120b-medium`
Median time to first token (gateway proxy) on Borg	6121 ms
Median tokens per second (end to end) on Borg	46.8 output tokens per second

Not published by the vendor: thinking modes, fusion roles.

What gpt-oss-120b is good for

gpt-oss-120b is the larger open-weight member of the gpt-oss family. The smaller 20b sibling favors low latency; unlike hosted GPT tiers, its weights are downloadable. It takes text input and produces text, with configurable reasoning effort and agent-style tool capabilities. Its Apache 2.0 license permits customization and local deployment. OpenAI does not publish a standard hosted API token rate for this model. It is useful when deployment control and access to weights matter more than a managed flagship endpoint. Borg includes this lane in its subscription; Borg Fusion can route to it, or you can select its lane directly for a fixed model choice.

private self-hosted reasoning
fine-tuning open weights
text-based tool workflows
configurable reasoning experiments

How Borg uses it

gpt-oss-120b is available as a subscription catalog lane in Borg. The catalog does not publish which Fusion profiles route to this model, so no routing role is claimed here. You can pick it directly by setting the OpenAI-compatible gateway model field to a listed lane id. Check the current catalog for availability before building a workflow around a lane.

model: "antigravity/gpt-oss-120b-medium"

Measured on Borg

Metric	Median	95th percentile	Samples
Time to first token (gateway proxy)	6121 ms	42094 ms	15830
Tokens per second (end to end)	46.8 output tokens per second	235 output tokens per second	92214
Request success rate	100%	—	21212

Time to first token (gateway proxy)

Lower is fasterFleet median 3971.5 ms

gpt-oss-120b6121 ms

Output tokens per second (whole turn)

Higher is fasterFleet median 31.9 tokens/s

gpt-oss-120b46.8 tokens/s

Time to first token (gateway proxy): time until the first streamed response event reaches the gateway; our closest available measure of time to first token
Tokens per second (end to end): tokens per second, end to end: provider-reported output tokens divided by the whole turn's elapsed time, including queueing and reasoning; not a streaming per-token rate

Share of requests Borg completed successfully, excluding requests the client cancelled.

Completion speed covers the whole turn including queueing and reasoning time; it is not a streaming tokens-per-second figure.

Measured on Borg: SearchAtlas AI Gateway production request logs. 30-day window ending 2026-10-08; 93142 requests. Minimum 100 samples per published request metric.

Benchmarks

Benchmark	Vendor-reported score	Source
AIME 2025 (high reasoning, no tools)	92.5	Vendor source
GPQA Diamond (high reasoning, no tools)	80.1	Vendor source
SWE-bench Verified (high reasoning)	62.4	Vendor source
MMLU (high reasoning)	90	Vendor source
Tau-Bench Retail (high reasoning)	67.8	Vendor source

Pricing

Included in Borg subscription. Vendor API list price: Not published.

Compare subscription and API pricing · Cost Savings Calculator

FAQ

What is gpt-oss-120b?

gpt-oss-120b is listed in Borg's subscription catalog as antigravity/gpt-oss-120b-medium. gpt-oss-120b is the larger open-weight member of the gpt-oss family.

Is gpt-oss-120b in Borg?

Yes. gpt-oss-120b is listed in Borg's subscription catalog under the lane antigravity/gpt-oss-120b-medium. You can choose the lane directly, or explore the available Fusion profiles. Availability and model details can change; check the current catalog before deployment.

What context window does gpt-oss-120b have?

The published context window is 131,072 tokens. Consult the cited sources for details and current limits. Vendor source

Is gpt-oss-120b good for coding?

Vendor-published evaluations appear in the benchmarks section; their settings may differ from your tasks. Its listed uses include private self-hosted reasoning, fine-tuning open weights, text-based tool workflows, configurable reasoning experiments. Try the lane for your own tasks, and compare the output with your requirements before choosing a default.

How fast is gpt-oss-120b on Borg?