Provider Capability and Model Catalog¶
DeltaLLM currently declares 20 provider presets and 79 curated model entries. Capabilities come from the runtime provider registry; model entries come from the validated catalogs packaged with DeltaLLM.
A capability means DeltaLLM has an implementation path for that configured model mode. It does not mean every upstream model supports every parameter. Confirm model-specific limits with the provider and test the exact deployment before production.
Capability matrix¶
| Provider | Protocol | Chat/text | Embeddings | Images | Speech | Transcription | Realtime WebSocket | Rerank | Curated models | Last verified |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek | OpenAI-compatible | Yes | — | — | — | — | — | — | 2 | 2026-09-09 |
| Z.ai | OpenAI-compatible | Yes | — | — | — | — | — | — | 3 | 2026-09-09 |
| Qwen (Alibaba Model Studio) | OpenAI-compatible | Yes | — | — | — | — | — | — | 2 | 2026-09-09 |
| Tencent TokenHub | OpenAI-compatible | Yes | — | — | — | — | — | — | 2 | 2026-09-09 |
| MiniMax | OpenAI-compatible | Yes | — | — | — | — | — | — | 2 | 2026-09-09 |
| OpenAI | OpenAI-compatible | Yes | Yes | Yes | Yes | Yes | Yes | — | 22 | 2026-04-01 |
| Anthropic | Anthropic-native | Yes | — | — | — | — | — | — | 3 | 2026-04-01 |
| Azure OpenAI | OpenAI-compatible | Yes | Yes | Yes | Yes | Yes | — | — | — | — |
| OpenRouter | OpenAI-compatible | Yes | Yes | Yes | — | — | — | — | — | — |
| Groq | OpenAI-compatible | Yes | Yes | — | Yes | Yes | — | — | 16 | 2026-04-01 |
| Together AI | OpenAI-compatible | Yes | Yes | Yes | — | — | — | — | — | — |
| Fireworks AI | OpenAI-compatible | Yes | Yes | Yes | — | — | — | — | 1 | 2026-04-01 |
| DeepInfra | OpenAI-compatible | Yes | Yes | Yes | — | — | — | — | — | — |
| Perplexity | OpenAI-compatible | Yes | — | — | — | — | — | — | — | — |
| Google Gemini | Native adapter | Yes | — | — | Yes | — | — | — | 20 | 2026-04-01 |
| Amazon Bedrock | Native adapter | Yes | — | — | — | — | — | — | 1 | 2026-04-01 |
| ElevenLabs | Native adapter | — | — | — | Yes | Yes | — | — | 5 | 2026-06-13 |
| vLLM | OpenAI-compatible | Yes | Yes | Yes | Yes | Yes | — | Yes | — | — |
| LM Studio | OpenAI-compatible | Yes | Yes | — | — | — | — | — | — | — |
| Ollama | OpenAI-compatible | Yes | Yes | — | — | — | — | — | — | — |
Preset endpoints¶
| Provider | Default API base | Protocol |
|---|---|---|
| DeepSeek | https://api.deepseek.com |
OpenAI-compatible |
| Z.ai | https://api.z.ai/api/paas/v4 |
OpenAI-compatible |
| Qwen (Alibaba Model Studio) | https://dashscope-intl.aliyuncs.com/compatible-mode/v1 |
OpenAI-compatible |
| Tencent TokenHub | https://tokenhub-intl.tencentcloudmaas.com/v1 |
OpenAI-compatible |
| MiniMax | https://api.minimax.io/v1 |
OpenAI-compatible |
| OpenAI | https://api.openai.com/v1 |
OpenAI-compatible |
| Anthropic | https://api.anthropic.com/v1 |
Anthropic-native |
| Azure OpenAI | https://{resource}.openai.azure.com/openai/v1 |
OpenAI-compatible |
| OpenRouter | https://openrouter.ai/api/v1 |
OpenAI-compatible |
| Groq | https://api.groq.com/openai/v1 |
OpenAI-compatible |
| Together AI | https://api.together.xyz/v1 |
OpenAI-compatible |
| Fireworks AI | https://api.fireworks.ai/inference/v1 |
OpenAI-compatible |
| DeepInfra | https://api.deepinfra.com/v1/openai |
OpenAI-compatible |
| Perplexity | https://api.perplexity.ai |
OpenAI-compatible |
| Google Gemini | https://generativelanguage.googleapis.com/v1beta |
Native adapter |
| Amazon Bedrock | https://bedrock-runtime.{region}.amazonaws.com |
Native adapter |
| ElevenLabs | https://api.elevenlabs.io/v1 |
Native adapter |
| vLLM | Operator supplied | OpenAI-compatible |
| LM Studio | Operator supplied | OpenAI-compatible |
| Ollama | Operator supplied | OpenAI-compatible |
Curated model catalogs¶
Catalog entries provide selectable model IDs and metadata; they do not restrict operators from configuring another valid upstream model ID.
Anthropic¶
Last verified: 2026-04-01 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
claude-opus-4-6 |
active | Chat/text | — |
claude-sonnet-4-6 |
active | Chat/text | — |
claude-haiku-4-5 |
active | Chat/text | — |
Sources:
Amazon Bedrock¶
Last verified: 2026-04-01 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
anthropic.claude-3-5-sonnet-20240620-v1:0 |
active | Chat/text | — |
Sources:
DeepSeek¶
Last verified: 2026-09-09 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
deepseek-v4-flash |
active | Chat/text | — |
deepseek-v4-pro |
active | Chat/text | — |
Sources:
ElevenLabs¶
Last verified: 2026-06-13 from mixed sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
eleven_v3 |
active | Speech | — |
eleven_multilingual_v2 |
active | Speech | — |
eleven_flash_v2_5 |
active | Speech | — |
scribe_v2 |
active | Transcription | — |
scribe_v1 |
deprecated | Transcription | — |
Sources:
Fireworks AI¶
Last verified: 2026-04-01 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
accounts/fireworks/models/llama-v3p1-405b-instruct |
active | Chat/text | — |
Sources:
Google Gemini¶
Last verified: 2026-04-01 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
gemini-3.1-pro-preview |
preview | Chat/text | — |
gemini-3-flash-preview |
preview | Chat/text | — |
gemini-3.1-flash-lite-preview |
preview | Chat/text | — |
gemini-2.5-pro |
active | Chat/text | — |
gemini-2.5-flash |
active | Chat/text | — |
gemini-2.5-flash-preview-09-2025 |
preview | Chat/text | — |
gemini-2.5-flash-lite |
active | Chat/text | — |
gemini-2.5-flash-lite-preview-09-2025 |
preview | Chat/text | — |
gemini-2.0-flash |
active | Chat/text | — |
gemini-2.0-flash-001 |
active | Chat/text | — |
gemini-2.0-flash-exp |
preview | Chat/text | — |
gemini-2.0-flash-lite |
active | Chat/text | — |
gemini-2.0-flash-lite-001 |
active | Chat/text | — |
gemini-3.1-flash-image-preview |
preview | Images | — |
gemini-3-pro-image-preview |
preview | Images | — |
gemini-2.5-flash-image |
active | Images | — |
gemini-2.0-flash-preview-image-generation |
preview | Images | — |
gemini-2.5-flash-preview-tts |
preview | Speech | — |
gemini-2.5-pro-preview-tts |
preview | Speech | — |
gemini-embedding-001 |
active | Embeddings | — |
Sources:
Groq¶
Last verified: 2026-04-01 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
llama-3.1-8b-instant |
active | Chat/text | — |
llama-3.3-70b-versatile |
active | Chat/text | — |
openai/gpt-oss-120b |
active | Chat/text | — |
openai/gpt-oss-20b |
active | Chat/text | — |
groq/compound |
active | Chat/text | — |
groq/compound-mini |
active | Chat/text | — |
meta-llama/llama-4-scout-17b-16e-instruct |
active | Chat/text | — |
openai/gpt-oss-safeguard-20b |
active | Chat/text | — |
qwen/qwen3-32b |
active | Chat/text | — |
moonshotai/kimi-k2-instruct-0905 |
active | Chat/text | — |
meta-llama/llama-prompt-guard-2-22m |
active | Chat/text | — |
meta-llama/llama-prompt-guard-2-86m |
active | Chat/text | — |
canopylabs/orpheus-arabic-saudi |
active | Speech | — |
canopylabs/orpheus-v1-english |
active | Speech | — |
whisper-large-v3 |
active | Transcription | — |
whisper-large-v3-turbo |
active | Transcription | — |
Sources:
MiniMax¶
Last verified: 2026-09-09 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
MiniMax-M3 |
active | Chat/text | — |
MiniMax-M2.7 |
active | Chat/text | — |
Sources:
OpenAI¶
Last verified: 2026-04-01 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
gpt-5.4 |
active | Chat/text | — |
gpt-5.4-mini |
active | Chat/text | — |
gpt-5.4-nano |
active | Chat/text | — |
gpt-5 |
active | Chat/text | — |
gpt-5-mini |
active | Chat/text | — |
gpt-5-nano |
active | Chat/text | — |
gpt-4.1 |
active | Chat/text | — |
gpt-4.1-mini |
active | Chat/text | — |
gpt-4o |
active | Chat/text | — |
gpt-4o-mini |
active | Chat/text | — |
gpt-4 |
legacy | Chat/text | — |
text-embedding-3-small |
active | Embeddings | — |
text-embedding-3-large |
active | Embeddings | — |
gpt-image-1.5 |
active | Images | — |
gpt-image-1-mini |
active | Images | — |
gpt-image-1 |
active | Images | — |
dall-e-3 |
legacy | Images | — |
tts-1 |
active | Speech | — |
tts-1-hd |
active | Speech | — |
gpt-4o-transcribe |
active | Transcription | — |
gpt-4o-mini-transcribe |
active | Transcription | — |
whisper-1 |
legacy | Transcription | — |
Sources:
Qwen (Alibaba Model Studio)¶
Last verified: 2026-09-09 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
qwen3.8-max |
active | Chat/text | — |
qwen3.8-flash |
active | Chat/text | — |
Sources:
Tencent TokenHub¶
Last verified: 2026-09-09 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
hy3 |
active | Chat/text | — |
hy4-preview |
preview | Chat/text | — |
Sources:
Z.ai¶
Last verified: 2026-09-09 from official_docs sources.
| Model ID | Status | Supported modes | Notes |
|---|---|---|---|
glm-5.3 |
active | Chat/text | — |
glm-5.3-flash |
active | Chat/text | — |
glm-5.2 |
active | Chat/text | — |
Sources: