gpt-oss-20b
OpenAI · Open weights · Released Aug 5, 2025 · openai/gpt-oss-20b
Best $0.018 in / $0.09 out via Darkbloom · fp8
131K contextToolsReasoningStructured outputPrompt cachingOpen weights
Available from 12 providersUSD / MTOK · PRICES 23 MINUTES AGO · SPEED 3 DAYS AGO
EST. MONTHLY COSTclick a row for detail · click headers to sort
| PROVIDER | INPUT | OUTPUT | CACHE | CONTEXT | QUANT | TTFT | TPS | SMOKE | UPTIME |
|---|---|---|---|---|---|---|---|---|---|
 Darkbloom | $0.018 | $0.09 | $0.009 | 131K | fp8 | ✕ | ✕ | — | 100.0% |
 AkashML | $0.02 | $0.10 | — | 131K | fp4 | 2.4s | 29 tok/s | — | 99.9% |
| DEDekaLLM | $0.029 | $0.14 | $0.029 | 131K | bf16 | 1.5s | 20 tok/s | — | 99.5% |
 CoreWeave | $0.03 | $0.13 | $0.03 | 131K | fp4 | 0.28s | 306 tok/s | 5/5 | 100.0% |
 DeepInfra | $0.03 | $0.14 | — | 131K | bf16 | 0.28s | 100 tok/s | 2/5 | 99.3% |
 Parasail | $0.03 | $0.15 | $0.02 | 131K | fp4 | 0.89s | 52 tok/s | 5/5 | 99.6% |
 Novita | $0.04 | $0.15 | — | 131K | fp4 | ✕ | ✕ | 4/5 | 100.0% |
 SiliconFlow | $0.04 | $0.18 | — | 131K | fp8 | 1.8s | 103 tok/s | 4/5 | 87.6% |
 Amazon Bedrock | $0.07 | $0.15 | — | 131K | — | 0.67s | 1667 tok/s | — | 100.0% |
| Amazon Bedrock | $0.07 | $0.15 | — | 131K | — | — | — | — | 100.0% |
 Google | $0.07 | $0.25 | — | 131K | — | 0.57s | 390 tok/s | — | 97.8% |
 Groq | $0.075 | $0.30 | $0.037 | 131K | — | 0.66s | 897 tok/s | — | 99.8% |
TTFT/TPS: median (P50) over trailing 504h · chat10k profile (~10K in / 200 out) · measured via OpenRouter routing · ✕ = probed, endpoint returned no measurable stream · SMOKE: of 5 programmatic capability probes passed (JSON schema, instruction following, tool call, long-context, benign compliance; retry-once) · screened via OpenRouter routing · v0.2 · methodology v1.1 · measured 3 days ago
SERVE THIS MODEL AND NOT LISTED? LIST YOUR INFERENCE →
Input price history · best listed price, observed
↓ 10% SINCE SEP 19, 2026
$0.02 · SEP 19, 2026$0.018 · TODAY
Pricing variants · USD / MTok in / out
STANDARD
$0.018 / $0.09
per MTok in / out
BATCH
$0.024 / $0.112
async batch
CACHE READ $0.009
Recent changesRSS ↗
Oct 5, 2026endpoint removed · primeintellect13 → 12
Oct 4, 2026endpoint added · primeintellect12 → 13
Sep 23, 2026endpoint removed · together13 → 12
Sep 19, 2026price decreased · input0.02 → 0.018
Sep 19, 2026price decreased · output0.1 → 0.09
Sep 17, 2026endpoint removed · phala14 → 13
Sep 12, 2026endpoint added · dekallm13 → 14
Sep 9, 2026endpoint removed · fireworks14 → 13
About
gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...
Specifications
Context window131,072
Max output32,768
Modalities intext
Modalities outtext
LicenseOpen weights
ReleasedAug 5, 2025
Knowledge cutoff2024-06-30
TokenizerGPT
WeightsHugging Face ↗
Related
Embed badge
[](https://modelindex.ai/models/openai/gpt-oss-20b)Sources
OpenRouter API23 minutes ago
AGGREGATORHugging Face23 minutes ago
COMMUNITYModelIndex benchmarks11 days ago
MEASUREDVerified 23 minutes ago · list prices, not negotiated rates.