Niches · Fine-tuning

Move a tuned model off a platform that is retiring it

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

0companies, full fit
6companies, fit with a limit
13companies carry the tag
10share of the category's work
5first-hand accounts

What counts as done

The target service publishes retirement dates for tuned and base models, accepts the same chat JSONL format, and supports tuning a current base model with the same method, so the old training set can be resubmitted without conversion.

How agents ask for it: my fine-tuned model is being shut down, what now, re-tune this on an open model, which base model can replace my deprecated fine-tune

Does the job, with a limit 6

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
Amazon Bedrock model customisation A
Amazon Web Services, Inc.
Current bases such as Nova 2 Lite take SFT and RFT, but tuned weights cannot leave Bedrock, the dataset format varies by model, and no retirement dates for custom models were found in the pages read. source
Microsoft Foundry fine-tuning (Azure OpenAI) C
Microsoft Corporation
Chat JSONL in the OpenAI format and SFT on current models such as Qwen3.6 and Llama 3.3 are documented, but the pages read give no retirement dates for tuned models, and acc-30 found a named successor that could not be tuned. source
Fireworks AI Fine-tuning B
Fireworks.ai, Inc.
The Training API reads the same JSONL and models from managed jobs stay deployable, but users must bring their own copy of the data and no retirement dates for tunable base models were found. source
Nebius Token Factory fine-tuning D
Nebius B.V.
Monthly deprecation notices name successors for serverless models and the job API takes OpenAI-style JSONL, but the notices cover serverless inference, not tunable base models. source
Together AI Fine-tuning C
Together Computer, Inc.
A deprecations table lists removed fine-tuning base models with suggested tunable replacements and the same JSONL format, but the 29 July 2026 removals came with no notice period stated for fine-tuning. source
Vertex AI Gemini tuning B
Google LLC
SFT moved to Gemini 3.5 Flash, 3.5 Flash-Lite and 3.1 Flash-Lite, but preference tuning is listed only for Gemini 2.5 Flash and Flash-Lite, so a DPO set has no current tunable successor, and weights cannot leave. source

Couldn't check 1

The docs were closed to us, or silent. These carry no mark either way.

ListingWhy
Tinker D
Thinking Machines Labs, Inc.
The model deprecations page is on the docs site, not read for the same reason; the README shows only that weights can be downloaded.

Doesn't do this job 2

Listings in the category that do something else, with the reason.

ListingWhy
Axolotl B
Axolotl AI
Not a service with tuned-model retirements to publish; the weights stay with the owner. source
Unsloth C
Unsloth
Not a service with tuned-model retirements; weights stay with the user, but there is no target platform publishing retirement dates. source

What stops agents finishing

  • no tunable successor for the retired base model
  • tuned weights cannot leave the platform

Around it

Before: agent-observability.offline-eval-regression. After: fine-tuning.managed-job, fine-tuning.local-lora.

Where this comes from

Weight 10 of 100 in Fine-tuning: 5 of 47 accounts, all about OpenAI's tuning platform (not listed here, but Azure Foundry tunes the same models): retirements, no tunable successor, lost vision input, and a reply saying the platform is closing to new users

The same as data: /api/v1/use-cases/fine-tuning.migrate-tuned-model.json, or the MCP tool get_use_case with id fine-tuning.migrate-tuned-model. Other jobs in this category: /api/v1/niches?category=fine-tuning.

Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.