KymaKyma
OpenAI

OpenAI

GPT-OSS 120B

#5 on Kyma

OpenAI's open-source 120B model, and a top-10 workhorse on Kyma — strong-tier quality at cheap-tier pricing, serving over 113M production tokens with a 100% success rate. Reach for it when you want dependable general intelligence and writing without flagship cost.

Modalities

Text → Text

Input

$0.0527 /1M

Output

$0.2565 /1M

Cached input

$0.00527 /1M90% off

Context

128K

Speed

medium

Performance

Live production data from real requests on Kyma — not synthetic benchmarks.

Rank

#5

of 87 active models

Tokens served

123.2M

all-time

Success rate

86.4%

last 7 days

Median throughput29 tok/s
Total requests15,648
Platform share5.2%
Tokens · last 15 daysJul 19 → Aug 2

Top apps using this model

1Python App4.0M tok
2OpenClaw2.4M tok
3Cline420.9K tok
4OpenAI SDK184.1K tok

Public apps sending the most traffic to this model — a signal of what real workloads it fits.

Pricing

Pay per token. Cached input is billed at 10% of the input rate.

$0.0527 /1M input$0.2565 /1M output
Hobby10 req/day · 2K in / 500 out
~$0.07/mo
Production1,000 req/day · 2K in / 500 out
~$7.01/mo
Scale20,000 req/day · 2K in / 500 out
~$140/mo
+ Estimate your workload
1,000
2,000
500
30%

Estimated monthly cost

$6.16

$0.2052 / day on GPT-OSS 120B

Same workload on:

GPT-5.6 Sol Pro$1,013+16348%
GPT-5.6 Sol$1,013+16348%

Estimates use list pricing with cached input billed at the 90%-discount rate. Actual bills depend on real token counts — every response includes its exact cost.

When to use GPT-OSS 120B

Updated 2026-06-10

Where this model earns its cost — and where it doesn't.

GPT-OSS 120B is OpenAI's open-source model at 120B parameters, built for general intelligence and writing. On Kyma it sits in the strong quality tier with cheap-tier pricing — the combination teams pick when they need broadly capable output across many requests and the bill matters.

It has earned that role in production: ranked #6 of 62 models by tokens served, with 5.7% of platform traffic across nearly 15,000 requests at a 100% success rate. Its biggest consumers are OpenClaw and OpenAI Python SDK integrations. Every call gets Kyma's automatic failover, and prompt caching is fully supported — repeated prompt prefixes bill at 10% of the input rate.

The 128K-token context window handles long documents and conversation history, while function calling and structured outputs make it usable inside agent pipelines, not just chat. It's text-in, text-out, with generations capped at 8K output tokens.

Writing and drafting

Its core strength — articles, summaries, rewrites, and editorial work at a price that holds up across high volume.

General assistants

A broadly capable default for chatbots and Q&A where you want strong-tier answers without flagship spend.

Agent pipelines

Function calling plus structured outputs keep it reliable inside multi-step agents — OpenClaw is already its top app by tokens on Kyma.

High-volume batch work

Cheap-tier pricing with prompt caching makes it economical for classification, extraction, and bulk text processing.

Long-context tasks

The 128K window fits substantial documents and long-running conversations in a single request.

Not ideal for: Image inputs (it's text-only), extended step-by-step reasoning modes (it has none), or single very long generations — output is capped at 8K tokens, and at a median 21 tokens/second it isn't the pick for latency-critical UX either.

How it compares

Against the peers people actually weigh it against.

SpecGPT-OSS 120BGPT-5.6 Sol ProGPT-5.6 Sol
Input /1M$0.0527$6.75$6.75
Output /1M$0.2565$40.50$40.50
Context128K1M1M
ToolsYesYesYes
ReasoningNoYesYes
Speedmediummediummedium

Quick start

Up and running in under two minutes.

  1. 1

    Create an API key

    Sign up and grab a key from the dashboard — $0.50 free credit, no card required.

    Get API key →
  2. 2

    Make your first request

    Drop in your key and send a chat completion — fully OpenAI-compatible.

    curl https://kymaapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-oss-120b",
    "messages": [
      {"role": "user", "content": "Explain prompt caching in one paragraph."}
    ]
  }'
  3. 3

    Stream responses

    Add "stream": true to receive tokens as they arrive.

    curl https://kymaapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-oss-120b",
    "stream": true,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

FAQ

Common questions about this model.

What is the context window of GPT-OSS 120B?

GPT-OSS 120B has a 128K-token context window — roughly 188 pages of text in a single request.

How much does the GPT-OSS 120B API cost?

$0.0527 per 1M input tokens and $0.2565 per 1M output tokens, with cached input at $0.00527/1M — a 90% discount on repeated prompt prefixes. No subscription; you pay only for what you use.

Does GPT-OSS 120B support function calling?

Yes — GPT-OSS 120B supports tool/function calling and structured outputs (JSON mode), so it works with agent frameworks out of the box.

How do I use GPT-OSS 120B?

Kyma is OpenAI-compatible: point your SDK's base URL at https://kymaapi.com/v1, use your Kyma API key, and set the model to gpt-oss-120b. Signing up is free and includes $0.50 of credit — no card required.

Is GPT-OSS 120B good for writing?

Yes — writing is what it's recommended for on Kyma, alongside general-purpose work. It produces strong-tier prose at cheap-tier rates, which is why it handles high-volume drafting and rewriting workloads well. For tasks that need extended chain-of-thought reasoning, pick a reasoning-capable model instead.

What throughput should I expect?

Production median is 21 tokens per second — a medium-speed model. That's comfortable for background jobs, batch processing, and agent steps, but if your product streams responses to users in real time, a fast-tier model will feel noticeably snappier.

Why use GPT-OSS 120B through Kyma?

One API key and one OpenAI-compatible endpoint covers this and every other model on the platform. You get automatic failover when a serving path degrades, prompt caching at a 90% discount on repeated prefixes, exact per-request cost in usage.cost, and $0.50 of free credit at signup — no card required.

Start with $0.50 free credit — no card required.Create account →

More models by OpenAI

See all 11 →
ModelContextInputOutput
OpenAIGPT-5.6 Terra1M$1.443$8.656
OpenAIGPT-5.6 Luna1M$0.1373$0.8231
OpenAIGPT-5.6 Luna Pro1M$0.135$0.81
OpenAIGPT-5.6 Terra Pro1M$1.35$8.10
OpenAIGPT-5.6 Sol1M$6.75$40.50
OpenAIGPT-5.6 Sol Pro1M$6.75$40.50
OpenAIGPT Realtime Translate—$0.0459 / min
OpenAIGPT Image 2—$0.081 / image