Skip to content

add glm-5, gpt-5-mini, deepseek-v3.2 models for vivgrid provider #915

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
rekram1-node merged 1 commit into from
Feb 15, 2026
Merged

add glm-5, gpt-5-mini, deepseek-v3.2 models for vivgrid provider #915

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 25 additions & 0 deletions providers/vivgrid/models/deepseek-v3.2.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,25 @@
name = "DeepSeek-V3.2"
family = "deepseek"
release_date = "2025-12-01"
last_updated = "2025-12-01"
attachment = false
reasoning = true
temperature = true
knowledge = "2024-07"
tool_call = true
open_weights = true
Copy link
Contributor

@rekram1-node rekram1-node Feb 14, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

isnt deepseek an interleaved reasoning model?

Copy link
Contributor Author

@fanweixiao fanweixiao Feb 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good question — DeepSeek-V3.2 is actually a hybrid / interleaved reasoning model rather than a dedicated reasoning-only model.

In most deployments (for example with vLLM), reasoning is not enabled by default and must be explicitly requested at runtime — e.g. by passing chat_template_kwargs.thinking=true in the request body. Without this, it behaves like a standard chat LLM.

In models.dev, the reasoning flag indicates that a model supports reasoning capabilities, not that it always operates in a dedicated reasoning mode. This is consistent with how other providers configure DeepSeek-V3.2.

In Vivgrid, we abstract these provider-specific details away — developers can enable reasoning using the same client interface they would use for GPT-5.2 , without needing to handle provider-specific request parameters.

e.g.

{
  "model": "DeepSeek-V3.2",
  "input": [
    {
      "role": "user",
      "content": "Explain why the sky is blue"
    }
  ],
  "reasoning": {
    "effort": "medium"
  },
}

Copy link
Contributor

@rekram1-node rekram1-node Feb 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ok


[cost]
input = 0.28
output = 0.42

[limit]
context = 128_000
output = 128_000

[modalities]
input = ["text"]
output = ["text"]

[provider]
npm = "@ai-sdk/openai-compatible"
29 changes: 29 additions & 0 deletions providers/vivgrid/models/glm-5.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
name = "GLM-5"
family = "glm"
release_date = "2026-02-12"
last_updated = "2026-02-12"
attachment = false
reasoning = true
temperature = true
tool_call = true
structured_output = true
open_weights = true

[interleaved]
field = "reasoning_content"

[cost]
input = 1.00
output = 3.20
cache_read = 0.2

[limit]
context = 202_752
output = 131_000

[modalities]
input = ["text"]
output = ["text"]

[provider]
npm = "@ai-sdk/openai-compatible"
26 changes: 26 additions & 0 deletions providers/vivgrid/models/gpt-5-mini.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
name = "GPT-5 Mini"
family = "gpt-mini"
release_date = "2025-08-07"
last_updated = "2025-08-07"
attachment = true
reasoning = true
temperature = false
knowledge = "2024-05-30"
tool_call = true
open_weights = false

[cost]
input = 0.25
output = 2.00
cache_read = 0.03

[limit]
context = 272_000
output = 128_000

[modalities]
input = ["text", "image"]
output = ["text"]

[provider]
npm = "@ai-sdk/openai-compatible"