Skip to content

feat: implement real LM Studio model discovery #32

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

Jump to bottom
Open
agustif wants to merge 459 commits into
base: dev
Choose a base branch
from
Open

feat: implement real LM Studio model discovery #32

agustif wants to merge 459 commits into from

Conversation

@agustif
Copy link
Owner

@agustif agustif commented Dec 19, 2025

Summary

  • Uses LM Studio API v0 for rich model metadata (type, arch, context)
  • Also calls OpenAI-compatible v1 API for additional fields (created, owned_by)
  • Merges data from both endpoints for complete model information
  • Sets accurate capabilities: vision from model.type, embeddings handling
  • Graceful fallback if either API fails
  • Zero guessing - only uses documented LM Studio API response fields

Technical Details

  • Calls /api/v0/models for comprehensive model information
  • Falls back to /v1/models for additional fields
  • Real context limits, architecture family, and modalities
  • Proper model type detection (llm, vlm, embeddings)

Testing

  • TypeScript compilation passes
  • OpenCode starts successfully with LM Studio integration
  • Graceful handling when LM Studio is offline

thdxr and others added 30 commits December 15, 2025 11:54
@thdxr
docs: restore desktop beta banner to homepage
46a35df
@adamdotdevin
fix: share page
b7a9cbf
@thdxr
docs: update header navigation to include desktop download
2c53abd
@actions-user
Update Nix flake.lock and hashes
284c045
@adamdotdevin
fix: image attachments in desktop
4236744
@adamdotdevin
fix: terminal in desktop
c4f6382
@adamdotdevin
fix: connect provider on homepage
7e49d0f
release: v1.0.156
cf05e6e
@thdxr
core: fix message caching for Anthropic models to improve response co…
7d1733c 
…nsistency
@actions-user
chore: format code
b1b1df8
@thdxr
ci: add app bundle target to fix macOS updater by generating .app.tar…
8b08e9c 
….gz files
@thdxr
ci: fix tauri updater version mismatch by checking out the release tag
8339031
release: v1.0.157
9c5f94b
@thdxr
core: fix server response handling to prevent connection timeouts
d5a506d
@actions-user
chore: format code
62cc532
@adamdotdevin
fix: allow for non-vcs projects in desktop
337a7e9
@justfortheloveof
tweak: prioritize fuzzysort results that start with user input (sst#5571
7abc2a9 
)
@adamdotdevin
fix: modal search
02b5e7d
release: v1.0.158
37e4c1e
@rekram1-node
add ability to set topK
81134cf
@actions-user
chore: format code
e555e89
@ariane-emory
fix: record shell mode in history (resolves sst#5454) (sst#5551)
6632987
@adamdotdevin
fix: missing event type (global)
300ec0e
@jayair
docs: make project names clickable links in ecosystem documentation
c3b3b13
@rekram1-node
tweak: add model flag support for agent create command
ca599ab
@adamdotdevin
fix: landing page CLS hero jump down
66f3e69
@adamdotdevin
fix: default steps expanded unless done
8010448
@adamdotdevin
fix: session nav on homepage
5388192
@adamdotdevin
fix: landing page CLS hero jump down
e16487b
@adamdotdevin
fix: keybinds for agent and model selection
4fd9a19
actions-user and others added 30 commits December 18, 2025 16:39
@actions-user
chore: format code
5cf8e54
@adamdotdevin
chore: cleanup
7bc47fb
@adamdotdevin
feat(desktop): shell mode
b787525
@adamdotdevin
fix(desktop): session ordered by most recent
b0aaf04
@adamdotdevin
fix(desktop): prompt history nav, optimistic prompt dup
268f37f
@adamdotdevin
fix(desktop): error styles
83d8a88
@adamdotdevin
fix(desktop): rendering shell mode messages
c868a40
@adamdotdevin
fix(desktop): checkbox render in safari fml
6f43d03
@elithrar
github: add OIDC_BASE_URL for custom GitHub App installs (sst#5756)
8b298a2
@actions-user
chore: format code
faeaafa
@adamdotdevin
fix(desktop): submit prompt
a6dd35d
@rekram1-node
rm interleaved thinking filter for certain kimi k2 thinking model pro…
9427f56 
…viders that were bugged
zen: cleanup headers
9998efd
@adamdotdevin
fix(desktop): don't show image button in shell mode
228b644
@adamdotdevin
fix(desktop): markdown styles
ee9f979
@OpeOginni
feat: add experimental support for Ty language server (sst#5575)
ab9ac7c
@actions-user
chore: format code
67cfd7f
@rekram1-node
chore: rm dead code
606cf3b
@rekram1-node
tweak: more retry cases
ecc5050
@rekram1-node
ci: handle case where generate.yml fails better
8d11df1
@rekram1-node
ci: fix file perm
1fe87b0
@rekram1-node
ci: fix generate
323ea10
@adamdotdevin
fix(desktop): smaller max-width when review open
af4087d
@adamdotdevin
chore: cleanup
15931fa
@adamdotdevin
fix(desktop): expanded states
0ebcaff
@adamdotdevin
fix: id
d57b963
@actions-user
chore: generate
c8de766
@actions-user
Update Nix flake.lock and hashes
a0ab3d9
release: v1.0.168
427157c
@agustif
feat: implement real LM Studio model discovery
b853ce4 
- Uses LM Studio API v0 for rich model metadata (type, arch, context)
- Also calls OpenAI-compatible v1 API for additional fields (created, owned_by)
- Merges data from both endpoints for complete model information
- Sets accurate capabilities: vision from model.type, embeddings handling
- Graceful fallback if either API fails
- Zero guessing - only uses documented LM Studio API response fields
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.