chore: sync model developers data#1795
Conversation
Greptile SummaryThis automated sync from the upstream
Confidence Score: 3/5The data-only sync is broadly safe, but the mistral-medium-latest alias now silently loses reasoning support, which could affect users who have that model configured. The mistral-medium-latest entry is replaced with a completely different model definition - reasoning goes from enabled to disabled, open_weights flips, and the cost drops by ~4x. Users who picked this alias for its reasoning capabilities will get a non-reasoning model without any warning after merge. The three removed OpenAI audio models (vs. deprecated elsewhere) are a minor inconsistency but could cause silent failures for any persisted user config that references those IDs. frontend/src/features/models/data/providers.json - specifically the mistral-medium-latest replacement and the removal of the three OpenAI audio models. Important Files Changed
Flowchart%%{init: {'theme': 'neutral'}}%%
flowchart TD
UP[Upstream all.json] -->|weekly sync workflow| PR[PR #1795]
PR --> A[reasoning_options added to existing models]
PR --> B[New models added: qwen3.7-plus, MiniMax-M3, gemini-2.5-flash-tts]
PR --> C[status deprecated added to Anthropic + Mistral models]
PR --> D[mistral-medium-latest replaced with new entry]
PR --> E[OpenAI audio models removed entirely]
D --> D1[Old: reasoning=true, open_weights=true, cost 1.5/7.5]
D --> D2[New: reasoning=false, open_weights=false, cost 0.4/2]
E --> E1[gpt-4o-mini-tts]
E --> E2[gpt-4o-transcribe]
E --> E3[whisper-1]
|
| "display_name": "Mistral Medium 3.1", | ||
| "type": "chat" | ||
| }, | ||
| { | ||
| "id": "mistral-medium-latest", | ||
| "name": "Mistral Medium (latest)", | ||
| "family": "mistral-medium", | ||
| "attachment": true, | ||
| "reasoning": { | ||
| "supported": false | ||
| }, | ||
| "tool_call": true, | ||
| "temperature": true, | ||
| "knowledge": "2025-05", | ||
| "release_date": "2025-08-12", | ||
| "last_updated": "2025-08-12", | ||
| "modalities": { | ||
| "input": [ | ||
| "text", | ||
| "image" | ||
| ], | ||
| "output": [ | ||
| "text" | ||
| ] | ||
| }, | ||
| "open_weights": false, | ||
| "limit": { | ||
| "context": 262144, | ||
| "output": 262144 | ||
| }, | ||
| "cost": { | ||
| "input": 0.4, | ||
| "output": 2 | ||
| }, | ||
| "display_name": "Mistral Medium (latest)", | ||
| "type": "chat" |
There was a problem hiding this comment.
The new mistral-medium-latest entry flips reasoning.supported from true to false, drops structured_output, changes open_weights from true to false, and cuts costs from 1.5/7.5 to 0.4/2 — meaning the "latest" alias now points to what is effectively a different, less capable model. Any user who selected this model expecting reasoning support will silently receive a non-reasoning model after this sync. Worth confirming this reflects the actual upstream API behaviour and that there is no downstream guard that would need updating.
| "display_name": "Devstral Small 2", | ||
| "type": "chat" | ||
| }, | ||
| { | ||
| "id": "devstral-latest", | ||
| "name": "Devstral 2", | ||
| "family": "devstral", | ||
| "attachment": false, | ||
| "reasoning": { | ||
| "supported": false | ||
| }, | ||
| "tool_call": true, | ||
| "temperature": true, | ||
| "knowledge": "2025-12", | ||
| "release_date": "2025-12-09", | ||
| "last_updated": "2025-12-09", | ||
| "modalities": { | ||
| "input": [ | ||
| "text" | ||
| ], | ||
| "output": [ | ||
| "text" | ||
| ] | ||
| }, | ||
| "open_weights": true, | ||
| "limit": { | ||
| "context": 262144, | ||
| "output": 262144 | ||
| }, | ||
| "status": "deprecated", | ||
| "cost": { | ||
| "input": 0.4, | ||
| "output": 2 | ||
| }, | ||
| "display_name": "Devstral 2", | ||
| "type": "chat" | ||
| }, | ||
| { | ||
| "id": "devstral-2512", | ||
| "name": "Devstral 2", |
There was a problem hiding this comment.
The new devstral-latest entry is added in this sync with "status": "deprecated" already set. A model alias introduced as deprecated from its very first appearance in the catalog is unusual — it suggests either the upstream data has a transient error, or this entry was deprecated in the same release cycle it was created. It is worth verifying whether this is intentional so as not to expose a dead alias in any UI model-picker.
Co-authored-by: looplj <231043908+looplj@users.noreply.github.com>
Automated sync of model developers data from upstream source.
Source: https://raw.githubusercontent.com/ThinkInAIXYZ/PublicProviderConf/refs/heads/dev/dist/all.json
This PR was automatically created by the weekly sync workflow.