Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
2 changes: 2 additions & 0 deletions frontend/public/logos/models/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,8 @@ The package declares an MIT license; the logos remain their owners' trademarks.
| mistral.svg | mistral-color.svg | Original color mark |
| nemotron.svg | nvidia-color.svg | Original color mark |

`llama.svg` is Meta's mark, shared by the Llama and Muse model families.

MiniCPM is not in that package. `minicpm.svg` uses the symbol (first path) from
[ModelBest's official MiniCPM SVG](https://www.modelbest.cn/modelbest/minicpm-en.aZf632Pf.svg),
linked by [its website](https://www.modelbest.cn/en/) and retrieved on 2026-09-17.
Expand Down
7 changes: 6 additions & 1 deletion frontend/src/catalog/index.ts
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@ export const QUANTS: Quant[] = [
{ id: 'fp16', label: 'FP16', bits: 16, format: 'Half precision, the unquantized weights' },
{ id: 'bf16', label: 'BF16', bits: 16, format: 'Brain float' },
{ id: 'q4_k_m', label: 'Q4_K_M', bits: 4, format: 'GGUF, used by llama.cpp and Ollama' },
{ id: 'q4_k_xl', label: 'Q4_K_XL', bits: 4, format: 'GGUF, Unsloth Dynamic Q4_K_XL' },
{ id: 'q4_0', label: 'Q4_0', bits: 4, format: 'GGUF' },
{ id: 'q5_k_m', label: 'Q5_K_M', bits: 5, format: 'GGUF' },
{ id: 'q6_k', label: 'Q6_K', bits: 6, format: 'GGUF' },
Expand Down Expand Up @@ -82,7 +83,11 @@ export const MODELS: Model[] = [
},
{
id: 'nemotron-3-5-lightning-30b-a3b', name: 'Nemotron 3.5 Lightning 30B A3B', family: 'Nemotron 3.5', params: '30B', architecture: 'moe', activeParams: '3B',
sourceUrl: 'https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16', logoUrl: '/logos/models/nemotron.svg', brandColor: '#8fc25c', quants: ['q4_k_m', 'q8_0', 'bf16'],
sourceUrl: 'https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16', logoUrl: '/logos/models/nemotron.svg', brandColor: '#8fc25c', quants: ['gptq-4bit', 'q4_k_m', 'q8_0', 'bf16'],
},
{
id: 'muse-glimmer-30b', name: 'Muse Glimmer 30B', family: 'Muse', params: '30B', architecture: 'dense',
sourceUrl: 'https://huggingface.co/facebook/Muse-Glimmer-30B', logoUrl: '/logos/models/llama.svg', brandColor: '#73b0ee', quants: ['q4_k_xl'],
},
{
id: 'qwen3-30b-a3b', name: 'Qwen3-30B-A3B', family: 'Qwen3', params: '30B', architecture: 'moe', activeParams: '3B',
Expand Down
18 changes: 18 additions & 0 deletions supabase/migrations/20260918034000_catalog_muse_nemotron.sql
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
begin;

-- Keep PR #31's catalog entries and model/quant boards in sync with the frontend.
insert into public.quants (id, label, bits, format) values
('q4_k_xl', 'Q4_K_XL', 4, 'GGUF, Unsloth Dynamic Q4_K_XL')
on conflict (id) do update set label = excluded.label, bits = excluded.bits, format = excluded.format;

insert into public.models (id, name, family, params, architecture, active_params, source_url, logo_url, brand_color) values
('muse-glimmer-30b', 'Muse Glimmer 30B', 'Muse', '30B', 'dense', null, 'https://huggingface.co/facebook/Muse-Glimmer-30B', '/logos/models/llama.svg', '#73b0ee'),
('nemotron-3-5-lightning-30b-a3b', 'Nemotron 3.5 Lightning 30B A3B', 'Nemotron 3.5', '30B', 'moe', '3B', 'https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16', '/logos/models/nemotron.svg', '#8fc25c')
on conflict (id) do update set name = excluded.name, family = excluded.family, params = excluded.params, architecture = excluded.architecture, active_params = excluded.active_params, source_url = excluded.source_url, logo_url = excluded.logo_url, brand_color = excluded.brand_color;

insert into public.model_quants (model_id, quant_id) values
('muse-glimmer-30b', 'q4_k_xl'),
('nemotron-3-5-lightning-30b-a3b', 'gptq-4bit')
on conflict (model_id, quant_id) do nothing;

commit;
Loading