GPT-OSS 20B
Comprehensive deployment profile and benchmark telemetry for GPT-OSS 20B. Self-hosting memory footprints, verified token economics, and direct API endpoints.
Verified Engineering Benchmarks
VRAM Requirements & Sizing
Formula: parameters × 2 bytes per parameter (FP16 precision).
Formula: parameters × 0.5 bytes per parameter (INT4/AWQ/GPTQ quantized). Theoretical minimum; actual may vary.
Deterministic VRAM math from entity graph. Green = fits in single GPU. Compare these GPUs side by side →
| GPU | VRAM | FP16 | FP8 | INT4 | Calculator Link |
|---|---|---|---|---|---|
| H100 SXM5 | 80 GB | OOM | ✓ fits | OOM | Pre-filled → |
| H200 | 141 GB | OOM | ✓ fits | OOM | Pre-filled → |
| B200 | 192 GB | OOM | ✓ fits | OOM | Pre-filled → |
| A100 80GB | 80 GB | OOM | ✓ fits | OOM | Pre-filled → |
| L40S | 48 GB | OOM | ✓ fits | OOM | Pre-filled → |
| RTX 4090 | 24 GB | OOM | OOM | OOM | Pre-filled → |
Hosting Topology & Recommended Clusters
Active API Providers & Free Tiers
| Provider | Input / 1M | Output / 1M | Speed | Free Tier Status |
|---|---|---|---|---|
| Groq | $0.07 | $0.30 | 1000 tok/s Provider PublishedMEDIUM SourceProvider Reported VerifiedSep 30, 2026 Value 1,000 tok/s Methodology Provider-published throughput benchmark. Actual throughput varies with context length, batch size, and concurrency. Verified Sep 30, 2026 (UTC) · Refresh cadence: DailySep 30, 2026 | ⚡ 30 RPM / 1,000 RPD (Free Plan) |
OpenAI SDK Drop-in
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.groq.com/openai/v1",
apiKey: process.env.GROQ_API_KEY,
});
const response = await client.chat.completions.create({
model: "gpt-oss-20b",
messages: [{ role: "user", content: "Hello" }],
});IDE Setup & Config
Generate configuration snippets for your preferred coding tool to run GPT-OSS 20B via a free backend.
~/.continue/config.json — No credit card required
🟠Settings → Models → Add Custom Model — No credit card
🟣ANTHROPIC_BASE_URL env var — No credit card required
🔵~/.config/opencode/opencode.json — No credit card required
🟢Local runtime — No API key required
🔧All tools + all backends — Copy-paste configs
Related Models & Providers
Connected entities in the OpenGPU Radar knowledge graph: sibling OpenAI models and API providers serving GPT-OSS 20B.
Educational Deep Dive
Explore Related
VRAM Requirements Guide
LearnLearn how VRAM requirements are calculated for different LLM models and quantization levels.
Quantization Explained
LearnUnderstand the trade-offs between FP16, FP8, and INT4 precision for LLM inference.
KV Cache Deep Dive
LearnMaster how key-value cache memory scales with context length and batch size.
Your LLM Deployment Journey
1. Understand VRAM Requirements
Learn how to calculate memory needs for your LLM
2. Select Compatible Hardware
Find GPUs that can run your selected model
3. Calculate Total Cost of Ownership
Compare self-hosting vs API costs for your use case
4. Deploy with vLLM
Step-by-step guide to production LLM deployment