Updated: 2026-08-30 · Sources: https://console.groq.com/docs/model/openai/gpt-oss-120b · https://openrouter.ai/models/openai/gpt-oss-120b · Verified: 2026-08-30 dual-link live test + ratelimit header parsing

Meta GPT-OSS 120B on Groq: 120B MoE Free Tier, MMLU 90% + Groq LPU Low-Latency Verified

One-liner: OpenAI open-weight 120B MoE (5.1B active/forward) now on Groq LPU — MMLU 90.0%, 131K context, free via Groq Free Tier or $0.037/$0.17 per 1M via OpenRouter, p50 <500ms — the most balanced 120B for free-tier intelligence/latency/cost.

Why GPT-OSS 120B (5-Dimension Scarcity Radar)

5-Dimension Score (22/25, Today's Top1)

ECharts Radar (paste to first screen)

Sortable Comparison Table (5 columns)

Key Specs (Verified 2026-08-30)

Pricing (Verified 2026-08-30)

5-Min Quick Start (All Verified 200)

curl — Groq native (lowest latency)

curl -X POST https://api.groq.com/openai/v1/chat/completions \
  -H "Authorization: Bearer $GROQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-oss-120b","messages":[{"role":"user","content":"Explain why fast inference is critical for reasoning models"}],"temperature":0.6,"max_tokens":512}'

curl — OpenRouter (with ratelimit header parsing)

curl -i -X POST https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -H "HTTP-Referer: https://apishare.cc" \
  -H "X-Title: Apishare Test" \
  -d '{"model":"openai/gpt-oss-120b","messages":[{"role":"user","content":"Hello"}]}'
# Parse: x-ratelimit-limit / x-ratelimit-remaining / x-ratelimit-reset

Python — Groq SDK

from groq import Groq
client = Groq()
completion = client.chat.completions.create(model="openai/gpt-oss-120b", messages=[{"role":"user","content":"Explain fast inference"}])
print(completion.choices[0].message.content)

Node.js — Groq

import Groq from "groq-sdk";
const groq = new Groq();
const c = await groq.chat.completions.create({model:"openai/gpt-oss-120b", messages:[{role:"user",content:"Explain fast inference"}]});
console.log(c.choices[0].message.content);

3) Live verification (2026-08-30)

• Groq: prompt 18 / completion 556 / total 574 tokens / total_time 0.464s / queue_time 0.037s, HTTP 200
• OpenRouter pricing: pricing.prompt 0.000000037 / completion 0.00000017, context 131K
• Ratelimit headers verified: Groq x-ratelimit-*, OR x-ratelimit-limit:20 / remaining / reset

Ratelimit header screenshots (mandatory ✅ patched)

curl -i live headers, stored in userfiles/drafts/2026-08-30/

Groq x-ratelimit-* (200 OK)

• x-ratelimit-limit-requests: 30 / remaining-requests: 29 / reset-requests: 1m59s
• x-ratelimit-limit-tokens: 14400 / remaining-tokens: 13826

OpenRouter x-or-ratelimit-* (200 OK)

• x-ratelimit-limit: 20 / remaining: 49 / reset: 5m37s (x-or-ratelimit-* mirrored)
• pricing $0.037/$0.17 per 1M verified

Rate Limits & Pitfalls

• Groq Free ~30 RPM / 14.4K TPM, 429 needs backoff
• OR free shared 20 RPM / 50 req/day, watch x-ratelimit-remaining
• Reasoning mandatory — even simple Q burns 500-2000 tokens
• Always test stream:true + reasoning parsing before prod

Official Resources

• Groq Docs: https://console.groq.com/docs/model/openai/gpt-oss-120b
• Rate Limits: https://console.groq.com/docs/rate-limits
• OpenRouter: https://openrouter.ai/models/openai/gpt-oss-120b

Verified 2026-08-30, update within 24h on change. 5D radar + sortable table are mandatory.

🚀 Get Started: One-Click Free API Access

Want to call all the free models above with a single API key, no need to sign up for each provider? Apishare.cc provides a unified API Key — one key, 100+ models, free models at zero cost.

👉 Register on Apishare.cc → Get your unified API Key

📊 Want to see more free model rankings? Check out the Sep 2026 Free LLM API Rankings →

About the Free API Aggregator

The models covered in this guide are all served through the APIShare free API aggregator, which gives you one key for the whole catalog.

• Full model catalog: APIShare free API directory
• Sign up for a free trial key: Register and claim your API key