gpt-oss-20b (batch)
- Input / output
- $0.018 / $0.09
- Providers
- 10
- Spread
- 5.8×
- 7 days
- +60.0%
- 30 days
- 0%
- Context
- 128K
Providers
| Provider | Input | Output | Blended | vs ref | Context | Uptime 1d | Price since |
|---|---|---|---|---|---|---|---|
AkashMLCheapest fp4 | $0.018 | $0.09 | $0.036 | ref | |||
| fp8 | $0.018 | $0.09 | $0.036 | ref | |||
DeepInfrabf16/flex bf16 | $0.024 | $0.112 | $0.046 | +28% | |||
| fp4 | $0.03 | $0.13 | $0.055 | +53% | |||
| bf16 | $0.029 | $0.14 | $0.0568 | +58% | |||
| bf16 | $0.03 | $0.14 | $0.0575 | +60% | |||
| fp4 | $0.03 | $0.15 | $0.06 | +67% | |||
| fp8 | $0.04 | $0.18 | $0.075 | +108% | |||
| $0.07 | $0.15 | $0.09 | +150% | ||||
Amazon Bedrockeu-west-1 | $0.07 | $0.15 | $0.09 | +150% | |||
| $0.075 | $0.30 | $0.131 | +265% | ||||
| fp4 | $0.186 | $0.278 | $0.209 | +481% |
- AkashMLCheapest$0.036$0.018 in · $0.09 outref vs ref128K ctx100% uptimefp4
- $0.036$0.018 in · $0.09 outref vs ref128K ctx100% uptimefp8
- bf16/flex$0.046$0.024 in · $0.112 out+28% vs ref128K ctx98.7% uptimebf16
- $0.055$0.03 in · $0.13 out+53% vs ref128K ctx99.7% uptimefp4
- $0.0568$0.029 in · $0.14 out+58% vs ref128K ctx94.5% uptimebf16
- $0.0575$0.03 in · $0.14 out+60% vs ref128K ctx99.9% uptimebf16
- $0.06$0.03 in · $0.15 out+67% vs ref128K ctx99.8% uptimefp4
- $0.075$0.04 in · $0.18 out+108% vs ref128K ctx89.1% uptimefp8
- $0.09$0.07 in · $0.15 out+150% vs ref128K ctx99.9% uptime
- eu-west-1$0.09$0.07 in · $0.15 out+150% vs ref128K ctx99.9% uptime
- $0.131$0.075 in · $0.30 out+265% vs ref128K ctx100% uptime
- $0.209$0.186 in · $0.278 out+481% vs ref128K ctx99.9% uptimefp4
Blended = (3 × input + output) / 4. Ref $0.036 = median of cheapest 20%.
Price history
USD per 1M tokens · each step is a price change
Prices in USD / 1M tokens.
Other pricing
| Offered by | Cache read |
|---|---|
| AkashML, Darkbloom | $0.009 |
| DeepInfra (bf16/flex) | $0.012 |
| CoreWeave | $0.03 |
| DekaLLM | $0.029 |
| DeepInfra | $0.015 |
| Parasail | $0.02 |
| Groq | $0.0375 |
| PrimeIntellect | $0.167 |
Changes
- PriceCache read price,$0.165→$0.167+1.2%
- PriceOutput price,$0.276→$0.278+0.7%
- PriceInput price,$0.184→$0.186+1.1%
- PriceCache read price,$0.157→$0.165+5.1%
- PriceOutput price,$0.262→$0.276+5.3%
- PriceInput price,$0.175→$0.184+5.1%
- PriceCache read price,$0.18→$0.157−12.8%
- PriceOutput price,$0.30→$0.262−12.7%
- PriceInput price,$0.20→$0.175−12.5%
- PriceCache read price,$0.024→$0.18+650%
- PriceOutput price,$0.04→$0.30+650%
- PriceInput price,$0.027→$0.20+641%
- PriceCache read price,$0.18→$0.024−86.7%
- PriceOutput price,$0.30→$0.04−86.7%
- PriceInput price,$0.20→$0.027−86.5%
About
The cheapest provider for gpt-oss-20b (batch) is AkashML at $0.036 per 1M tokens (blended), charging $0.018 per 1M input and $0.09 per 1M output tokens. 10 providers serve it; the most expensive charges 5.8× the typical price. Context window: 128K tokens. Prices tracked since Oct 3, 2026.
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...
- Released
- Aug 5, 2025
- First tracked
- Oct 3, 2026
- Canonical slug
- openai/gpt-oss-20b
- Weights
- openai/gpt-oss-20b