← Back to models

gpt-oss-120b

OpenAI · Mixture-of-Experts (MoE) · License: Apache-2.0

OpenAI's first open-weight release since GPT-2. Designed to run efficiently — only ~5.1B of its 117B parameters activate per token — while still needing its full weight set in memory.

Memory requirement — the visual calculator

In real deployments, memory needs also depend on numeric precision (e.g. FP16 vs INT8 vs INT4), which changes bytes-per-parameter. This store uses a simplified, transparent formula so beginners can follow the logic: (parameters in billions × 1 GB) + 20% overhead for the runtime, active requests, and context memory.

Model weights
117 GB
+ 20% overhead
23.4 GB
= Total required
140.4 GB

Calculation: 117B parameters × 1 GB = 117 GB, plus 20% (23.4 GB) = 140.4 GB total The dedicated memory built into the graphics card. An AI model's entire set of "weights" (its learned parameters) must fit into this memory before the card can run it. If a model needs more VRAM than one card has, you need multiple cards working together, or a card with more memory. needed. Note: even though only 5.1B parameters ( A model design where only a fraction of the model's parameters ("active parameters") are used to process any single piece of text, even though the full model is much larger. This makes the model faster to run than its total size suggests, but you still need enough memory to hold the entire model, not just the active part. ) are active per token, the full 117B must still be loaded into memory.

Cheapest way to run this model today
3× NVIDIA RTX PRO 5000 Blackwell (144 GB total VRAM, ~$19,416). See the full recommendation →

Leaderboard

Listed on the LMArena text leaderboard and LMArena WebDev leaderboard (added Aug 2025).

Sources