SunsetDiff

Nebius model shutdowns on November 9, 2026

The dataset has 11 Nebius ids with shutdown date 2026-11-09. The November notice removes them from Token Factory Serverless on that date. Each row stores one recommended replacement from that table. The notice says Dedicated Endpoints are unaffected, and Token Factory does not automatically reroute requests. 3 August 31, 2026 ids store a replacement that is itself in this November list.

Last reviewed . Publisher: SunsetDiff.

Steps

  1. Read the ids and the replacement. The first table is every Nebius id in the dataset whose shutdown date is 2026-11-09. The replacement cell is the recommended replacement stored from the November notice. Open the model page before you change a call.
  2. Check Dedicated Endpoints separately. The notice says this change only affects Serverless usage. A Dedicated Endpoint for the same model is unaffected. Token Factory does not automatically reroute a Serverless request, so the old id will not start calling the new one for you.
  3. Update model and base_url. The notice says to update the model and base_url values, then verify the results. The quickstart uses base_url https://api.tokenfactory.nebius.com/v1/. Put the replacement id from the same row in the model field.
  4. Test before the date. Call the replacement with the prompts you send now. Do that before November 9, 2026. If an August migration already moved you onto one of these November ids, follow that row's replacement one more step.

Models shutting down

These rows are every Nebius model in the dataset with shutdown date 2026-11-09. The model id links to its SunsetDiff model page. Deprecation date, shutdown date, and replacement are the fields stored on that row. Not published means the field is empty. These rows do not store a deprecation_date.

The November notice says: "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." It also says: "Affected Serverless models will no longer be available via API or Playground." And: "Dedicated Endpoints for these models will be unaffected by this change."

The same page says: "This change only affects Serverless usage." "Dedicated Endpoints are unaffected." "Token Factory does not automatically reroute requests from deprecated models." "Choose a replacement based on your workload, then test before migrating traffic." The replacement column is the Recommended Replacement cell on that page.

11 Nebius model ids with shutdown date 2026-11-09.
Model idDeprecation dateShutdown dateRecommended replacement
deepseek-ai/DeepSeek-V4-Flash-0731Not published2026-11-09deepseek-ai/DeepSeek-V4.1-Flash-1017
deepseek-ai/DeepSeek-V4-ProNot published2026-11-09deepseek-ai/DeepSeek-V4-Pro-0813
deepseek-ai/DeepSeek-V4.1-FlashNot published2026-11-09deepseek-ai/DeepSeek-V4.1-Flash-1017
moonshotai/Kimi-K2.6Not published2026-11-09moonshotai/Kimi-K3
moonshotai/Kimi-K2.7-CodeNot published2026-11-09moonshotai/Kimi-K3
nvidia/nemotron-3-super-120b-a12bNot published2026-11-09nvidia/Nemotron-3.5-Super-VL
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3BNot published2026-11-09nvidia/Nemotron-3_5-Lightning
openbmb/MiniCPM-V-4_5Not published2026-11-09nvidia/Nemotron-3.5-Super-VL
Qwen/Qwen3-235B-A22B-Instruct-2507Not published2026-11-09Qwen/Qwen3-30B-A3B-Instruct-2507
Qwen/Qwen3.5-397B-A17BNot published2026-11-09Qwen/Qwen3.8-27B
zai-org/GLM-5.1Not published2026-11-09zai-org/GLM-5.3

Source: Nebius November 2026 deprecations , checked November 1, 2026.

August replacements that shut down in November

These August 31, 2026 rows store a replacement that is also in the November table. The August notice is still the source for the August replacement. The November notice is the source for the next id. SunsetDiff does not replace the August row's stored replacement with the November one.

An app that already moved from the August id to that replacement still has to change the model field again. The November notice says requests are not rerouted.

August 31, 2026 Nebius rows whose stored replacement shuts down on 2026-11-09.
August model idAugust shutdown dateStored replacementNovember recommended replacement
deepseek-ai/DeepSeek-V4-Flash2026-08-31deepseek-ai/DeepSeek-V4-Flash-0731deepseek-ai/DeepSeek-V4.1-Flash-1017
nvidia/Llama-3_1-Nemotron-Ultra-253B-v12026-08-31nvidia/nemotron-3-super-120b-a12bnvidia/Nemotron-3.5-Super-VL
Qwen/Qwen3-Next-80B-A3B-Thinking2026-08-31Qwen/Qwen3.5-397B-A17BQwen/Qwen3.8-27B

Sources: Nebius August 2026 deprecations , Nebius November 2026 deprecations , checked November 1, 2026.

What breaks

On November 9, 2026 these ids reach the shutdown date stored in the dataset. The notice says you will not be able to select them in the Playground or send API requests using their model IDs. It does not print an HTTP status.

It also says: "Token Factory does not automatically reroute requests from deprecated models." A request that still sends the old id is not forwarded to the replacement. A Dedicated Endpoint is a different case. The notice says: "Dedicated Endpoints are unaffected."

Source: Nebius November 2026 deprecations , checked November 1, 2026.

Find the model ids in your code

Search the repository for the ids in the November table, then for any August id in the second table. A short search for DeepSeek or Qwen will hit ids that are not in this shutdown. Read the full id. How to find every model ID in your codebase covers SDK argument names. The scan page builds a search from dataset ids that have a shutdown date.

The notice says: "Update the model and base_url values in your API requests, then verify the results." The quickstart sets base_url to https://api.tokenfactory.nebius.com/v1/ and posts to https://api.tokenfactory.nebius.com/v1/chat/completions. Its sample model is deepseek-ai/DeepSeek-R1-0528. That id is not in this table.

ripgrep

rg -n -w -e 'deepseek-ai/DeepSeek-V4-Flash-0731' -e 'deepseek-ai/DeepSeek-V4-Pro' -e 'deepseek-ai/DeepSeek-V4.1-Flash' -e 'moonshotai/Kimi-K2.6' -e 'moonshotai/Kimi-K2.7-Code' -e 'nvidia/nemotron-3-super-120b-a12b' -e 'nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B' -e 'openbmb/MiniCPM-V-4_5' -e 'Qwen/Qwen3-235B-A22B-Instruct-2507' -e 'Qwen/Qwen3.5-397B-A17B' -e 'zai-org/GLM-5.1' -e 'deepseek-ai/DeepSeek-V4-Flash' -e 'nvidia/Llama-3_1-Nemotron-Ultra-253B-v1' -e 'Qwen/Qwen3-Next-80B-A3B-Thinking'

Sources: Nebius November 2026 deprecations , Nebius quickstart , checked November 1, 2026.

Test the replacement before the cutoff

The notice says: "Choose a replacement based on your workload, then test before migrating traffic." Then: "Update the model and base_url values in your API requests, then verify the results." Call the replacement id from the same row with the prompts you send now. Do that before November 9, 2026.

The pre-migration test checklist is the longer list: evals, rejected fields, token limits, tools, price, and latency. These Nebius rows do not store a price.

OpenAI-compatible client, model field only

from openai import OpenAI
client = OpenAI(base_url="https://api.tokenfactory.nebius.com/v1/")
response = client.chat.completions.create(
    model="zai-org/GLM-5.3",
    messages=[{"role": "user", "content": "Hello"}],
)

Sources: Nebius November 2026 deprecations , Nebius quickstart , checked November 1, 2026.

Related SunsetDiff pages

All guides

Questions

Which Nebius ids shut down on November 9, 2026?

The dataset lists deepseek-ai/DeepSeek-V4-Flash-0731, deepseek-ai/DeepSeek-V4-Pro, deepseek-ai/DeepSeek-V4.1-Flash, moonshotai/Kimi-K2.6, moonshotai/Kimi-K2.7-Code, nvidia/nemotron-3-super-120b-a12b, nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B, openbmb/MiniCPM-V-4_5, Qwen/Qwen3-235B-A22B-Instruct-2507, Qwen/Qwen3.5-397B-A17B, zai-org/GLM-5.1 with shutdown date 2026-11-09. The notice says: "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."

What is the recommended replacement?

The recommended replacement is the official replacement stored on each row. deepseek-ai/DeepSeek-V4-Flash-0731: deepseek-ai/DeepSeek-V4.1-Flash-1017. deepseek-ai/DeepSeek-V4-Pro: deepseek-ai/DeepSeek-V4-Pro-0813. deepseek-ai/DeepSeek-V4.1-Flash: deepseek-ai/DeepSeek-V4.1-Flash-1017. moonshotai/Kimi-K2.6: moonshotai/Kimi-K3. moonshotai/Kimi-K2.7-Code: moonshotai/Kimi-K3. nvidia/nemotron-3-super-120b-a12b: nvidia/Nemotron-3.5-Super-VL. nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B: nvidia/Nemotron-3_5-Lightning. openbmb/MiniCPM-V-4_5: nvidia/Nemotron-3.5-Super-VL. Qwen/Qwen3-235B-A22B-Instruct-2507: Qwen/Qwen3-30B-A3B-Instruct-2507. Qwen/Qwen3.5-397B-A17B: Qwen/Qwen3.8-27B. zai-org/GLM-5.1: zai-org/GLM-5.3.

Will Token Factory reroute the old id?

The notice says: "Token Factory does not automatically reroute requests from deprecated models." You change the model field. It also says: "This change only affects Serverless usage." And: "Dedicated Endpoints are unaffected."

What does the API host look like?

The quickstart sets base_url to https://api.tokenfactory.nebius.com/v1/. The notice says: "Update the model and base_url values in your API requests, then verify the results."

What if I already migrated in August?

These August ids store a replacement that shuts down on 2026-11-09: deepseek-ai/DeepSeek-V4-Flash to deepseek-ai/DeepSeek-V4-Flash-0731. nvidia/Llama-3_1-Nemotron-Ultra-253B-v1 to nvidia/nemotron-3-super-120b-a12b. Qwen/Qwen3-Next-80B-A3B-Thinking to Qwen/Qwen3.5-397B-A17B. The November table names the next replacement. The August row keeps the August replacement.

More guides

Deprecation alerts

We email you before models shut down. Unsubscribe anytime.

Deprecation alert signup
Providers (optional)