gpt-oss-20b
Shutting down. gpt-oss-20b shuts down on Google Vertex AI on Oct 21, 2026. OpenAI names Qwen3.5-9B as the replacement.
7 days- Next shutdown
- 7 daysOct 21, 2026 · Google Vertex AI
- Arena Text
- 1,317#111 of 128 · Arena
- GPQA Diamond
- 60.8%#82 of 108 · Epoch AI
- List price · $ per 1M
- $0.07 in $0.30 outAmazon Bedrock
- Context window
- 131.1KMax output 131.1K tokens
Dates on every platform
| Platform | Model ID and status | Deprecated | Shuts down | Replacement |
|---|---|---|---|---|
| Amazon BedrockAmazon Web Services: Amazon Bedrock ListFoundationModels, us-east-1 (opens in a new tab) · checked Oct 14, 2026 | openai.gpt-oss-20b-1:0Active · also openai.gpt-oss-20b, us-gov.openai.gpt-oss-20b-1:0 | – | None set | – |
| Google Vertex AIGoogle Cloud: Open model deprecations (opens in a new tab) · checked Oct 14, 2026 | gpt-oss-20b-maasDeprecated | 7 days | – | |
| Together AITogether AI: Changelog (opens in a new tab) · checked Oct 14, 2026 | openai/gpt-oss-20bShut down | – | Shut down | Qwen3.5-9B |
| Fireworks AIFireworks AI: OpenAI gpt-oss-20b model page (opens in a new tab) · checked Oct 14, 2026 | accounts/fireworks/models/gpt-oss-20bShut down | – | Shut down | gpt-oss-120b |
Google suggests self-deploying the model from Model Garden or moving to another managed endpoint.
Together announced removal for 14 September 2026; the deprecation history records it on 15 September 2026.
Removed from serverless only; dedicated deployments of this model are unaffected.
What breaks when you move off gpt-oss-20b
- Availability
API requests to the retired model IDs fail after the retirement date.
DeepSeek-R1, DeepSeek-V3.2, DeepSeek-V3.1, GLM-5, GLM-4.7, gpt-oss-20b, Kimi K2 Thinking, Llama 3.3 70B Instruct, MiniMax-M2, Qwen3-235B-A22B-Instruct-2507, Qwen3-Coder-480B-A35B-Instruct, Qwen3-Next-80B-A3B-Instruct, Qwen3 Next 80B A3B Thinking · Google Vertex AI ·
- Availability
- Availability
Intelligence
| Benchmark | Publisher | Score | Rank | Published |
|---|---|---|---|---|
| Arena Text | Arena | 1,317 ± 3 | #111 of 128 | Oct 8, 2026 |
| GPQA Diamond | Epoch AI | 60.8% ± 2.5% | #82 of 108 | Aug 27, 2026 |
| OTIS Mock AIME 2024-2025 | Epoch AI | 65.3% ± 5.9% | #75 of 105 | Aug 27, 2026 |
Arena (arena.ai), https://arena.ai/leaderboard
Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource].
List prices
| Platform | Input | Cached | Output | In effect | Source |
|---|---|---|---|---|---|
Amazon Bedrockopenai.gpt-oss-20b-1:0 | $0.07 | – | $0.30 | Oct 1, 2026Current · Standard on-demand tier in us-east-1. | Amazon Web Services: AWS Price List API: AmazonBedrock, us-east-1 (opens in a new tab) |
Change history
ShutdownDeepSeek, GLM, Kimi, Llama, MiniMax, Qwen and GPT OSS endpoints retire on Agent PlatformGoogle Vertex AI1 break
Google retires its managed endpoints for DeepSeek, GLM, GPT OSS 20B, Kimi K2 Thinking, Llama 3.3 70B, MiniMax M2 and Qwen3 models on Gemini Enterprise Agent Platform. Google suggests self-deploying them from Model Garden.
- Platform
- Google Vertex AI
- Models
- DeepSeek-R1, DeepSeek-V3.2, DeepSeek-V3.1, GLM-5, GLM-4.7, gpt-oss-20b, Kimi K2 Thinking, Llama 3.3 70B Instruct, MiniMax-M2, Qwen3-235B-A22B-Instruct-2507, Qwen3-Coder-480B-A35B-Instruct, Qwen3-Next-80B-A3B-Instruct, Qwen3 Next 80B A3B Thinking
- Announced
- AvailabilityAPI requests to the retired model IDs fail after the retirement date.
Source: Google Cloud, Gemini Enterprise Agent Platform release notes (opens in a new tab); Google Cloud, Open model deprecations (opens in a new tab)
Shutdowngpt-oss-20b and Gemma 4 31B leave Together serverlessTogether AI1 break
Together removed openai/gpt-oss-20b (replacement Qwen3.5 9B) and google/gemma-4-31B-it (replacement GLM-5.3 Flash) from serverless. The removal was announced for 14 September 2026 and is recorded on 15 September 2026.
- Platform
- Together AI
- Models
- gpt-oss-20b, Gemma 4 31B
- Announced
- AvailabilityRequests to the removed serverless model ID fail. The model can still run on an on-demand dedicated endpoint, billed per minute.
Source: Together AI, Deprecations (opens in a new tab); Together AI, Changelog (opens in a new tab)
ShutdownFireworks removes MiniMax M2.7, GPT OSS 20B, DeepSeek V4 Pro and fast Kimi variants from serverlessFireworks AI1 break
MiniMax M2.7, GPT OSS 20B and the original DeepSeek V4 Pro left serverless, along with the Turbo and Fast variants of Kimi K2.6 and the Fast variant of Kimi K2.7 Code. Fireworks named MiniMax M3, GPT OSS 120B, DeepSeek V4 Pro (0813) and the standard Kimi modes as migration targets.
- Platform
- Fireworks AI
- AvailabilityServerless requests to these model IDs stop working after the date. Dedicated deployments are unaffected.