Groq LPUOpen-WeightVerified 2026-11-12
GPT-OSS 120B (Groq) API Pricing & Cost Rules
API Model ID: openai/gpt-oss-120b · Context Window: 131,072 tokens · Max Output: 65,536 tokens
Context Tiers & Token Rate Card (USD per 1M Tokens)
| Tier | Std Input | Std Output | Cache Read | Write (Short) | Write (Long) |
|---|---|---|---|---|---|
| Standard (<= 131K) | $0.15 | $0.6 | $0.075 | $0.15 | N/A |
Prompt Caching Rules
Supported: Yes (automatic)
Minimum Cacheable Prefix: None / Automatic
Hourly Cache Storage: $0
Automatic volatile memory cache (2h inactivity expiry), 50% off input ($0.075/1M), zero write surcharge. Supported ONLY on 3 Groq models (gpt-oss-120b, gpt-oss-20b, gpt-oss-safeguard-20b).
Batch API & Hosting Notes
Batch Supported: Yes (50% discount · 24h-7d)
Caching in Batch: Disabled in Batch mode
50% discount ($0.075/$0.30). Groq explicitly states Batch API does NOT stack with prompt caching.
~500 tokens/sec throughput on Groq LPU; 250K TPM / 1K RPM Developer tier.