Skip to content

Update synthetic.new models: promote MiniMax-M2.5, add GLM-4.7-Flash#1018

Merged
rekram1-node merged 3 commits intoanomalyco:devfrom
Sewer56:add-synthetic-missing-models
Mar 11, 2026
Merged

Update synthetic.new models: promote MiniMax-M2.5, add GLM-4.7-Flash#1018
rekram1-node merged 3 commits intoanomalyco:devfrom
Sewer56:add-synthetic-missing-models

Conversation

@Sewer56
Copy link
Copy Markdown
Contributor

@Sewer56 Sewer56 commented Feb 24, 2026 •
edited
Loading

Summary

Updates synthetic.new model configurations based on API changes:

  • hf:MiniMaxAI/MiniMax-M2.5 - promoted from beta to stable
  • hf:zai-org/GLM-4.7-Flash - newly added
  • hf:Qwen/Qwen3.5-397B-A17B - removed (retired)
    • Was in earlier version of this PR

Changes

All specs sourced from https://api.synthetic.new/openai/v1/models:

GLM-4.7-Flash (new)

  • Context: 196,608 / Output: 65,536
  • Pricing: $0.06/M input, $0.40/M output, $0.06/M cache read
  • Capabilities: reasoning, tool_call, structured_output, interleaved reasoning

MiniMax-M2.5

  • Removed beta status (now stable)

Checklist

  • Configuration validated with bun validate
  • Specs match API response from synthetic.new

@Sewer56
Add missing synthetic.new models
eee3303 
Add configuration for hf:Qwen/Qwen3.5-397B-A17B and hf:MiniMaxAI/MiniMax-M2.5
to the synthetic provider, based on API specs from synthetic.new.

Note: API reports image support but these models may not natively support
images (likely rerouted/proxied through vision-capable infrastructure).
@Sewer56 Sewer56 changed the title Add missing synthetic.new models: Qwen3.5-397B-A17B and MiniMax-M2.5 Add beta synthetic.new models: Qwen3.5-397B-A17B and MiniMax-M2.5 Feb 24, 2026
@Sewer56 Sewer56 marked this pull request as ready for review February 24, 2026 09:24
@raedkit
Copy link
Copy Markdown

raedkit commented Feb 24, 2026

Hi, interested in this PR (was planning to start it and fortunately checked before).
Any chance to get it merged today ? Thanks in advance.

@mugnimaestra
Copy link
Copy Markdown
Contributor

mugnimaestra commented Feb 24, 2026

https://qwen.ai/blog?id=qwen3.5

Qwen3.5-397B-A17B does natively support image modalities

@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Feb 25, 2026

This should be good to merge now.
I confirmed with provider; MM2.5 showing image was a mistake.
And the Qwen3.5 model had the comment removed as it truly is also image input.

@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Feb 28, 2026

@rekram1-node

@rekram1-node
Copy link
Copy Markdown
Collaborator

rekram1-node commented Mar 7, 2026

Was on vacation, just got back.

@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Mar 7, 2026

Qwen got retired earlier today due to perf issues.
Will need to update this. Because GLM4.7 Flash was also added.
(All happened a few hours back)

@Sewer56 Sewer56 changed the title Add beta synthetic.new models: Qwen3.5-397B-A17B and MiniMax-M2.5 Update synthetic.new models: retire Qwen3.5, promote MiniMax-M2.5, add GLM-4.7-Flash Mar 8, 2026
@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Mar 8, 2026

Okay sorted. CC @rekram1-node

@Sewer56 Sewer56 force-pushed the add-synthetic-missing-models branch from d1fd4b5 to 7a02946 Compare March 8, 2026 22:56
@Sewer56 Sewer56 changed the title Update synthetic.new models: retire Qwen3.5, promote MiniMax-M2.5, add GLM-4.7-Flash Update synthetic.new models: promote MiniMax-M2.5, add GLM-4.7-Flash Mar 8, 2026
@rekram1-node rekram1-node merged commit dc98a32 into anomalyco:dev Mar 11, 2026
1 check passed
@rekram1-node
Copy link
Copy Markdown
Collaborator

rekram1-node commented Mar 11, 2026

Thanks!

@themarcel
Copy link
Copy Markdown

themarcel commented Mar 30, 2026

Is anyone checking a fix for the Qwen3.5 perf issues?

@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Mar 30, 2026

Qwen3.5 was discontinued a while back at Synthetic, many weeks ago-

@themarcel
Copy link
Copy Markdown

themarcel commented Mar 31, 2026

Qwen3.5 was discontinued a while back at Synthetic, many weeks ago-

I use Synthetic and it is available and included in the models hf:Qwen/Qwen3.5-397B-A17B in https://dev.synthetic.new/docs/api/models.

I also added it in my fork of models.dev and it works with the same code that this PR had.

@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Mar 31, 2026

Ah, it was made proxied.
I thought the model was yanked entirely.

@themarcel
Copy link
Copy Markdown

themarcel commented Mar 31, 2026 •
edited
Loading

Ah, it was made proxied. I thought the model was yanked entirely.

Should it be re-added then? I can create a PR if that is the case.

@Sewer56
Copy link
Copy Markdown
Contributor Author

Sewer56 commented Mar 31, 2026

Yeah go for it.

@themarcel themarcel mentioned this pull request Mar 31, 2026
Merged
@themarcel
Copy link
Copy Markdown

themarcel commented Mar 31, 2026

Yeah go for it.

Done in #1305

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

1 more reviewer

@rudironsoni rudironsoni rudironsoni approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@Sewer56 @raedkit @mugnimaestra @rekram1-node @themarcel @rudironsoni