Supported Models
Models from all major AI providers, unified under a single model name. This page is auto-rendered from data/models.yaml, so the list always stays in sync with the data file.
How to add a model
lurus-docs/data/models.yaml. After you push, CI builds and updates automatically.Model List
DeepSeek
Everyday chat, code generation, Chinese writing, math reasoning
| Model | Context Length | Pricing | Status | Tags |
|---|---|---|---|---|
deepseek-chat | 64K | ¥1/M tokens | Available | ChatChinese |
deepseek-reasoner | 64K | ¥2/M tokens | Available | ReasoningMathCode |
deepseek-v3-2-251201 | 64K | ¥2/M tokens | Available | Latest |
Google Gemini
Long-document analysis, image understanding, multi-turn dialogue
| Model | Context Length | Pricing | Status | Tags |
|---|---|---|---|---|
gemini-3-pro-preview | 1M | ¥5/M tokens | Beta | MultimodalLong Context |
gemini-3-flash-preview | 1M | ¥2/M tokens | Beta | FastLong Context |
gemini-3-pro-image-preview | 1M | ¥5/M tokens | Beta | Image |
OpenAI
Complex tasks, specialized domains, English-first
| Model | Context Length | Pricing | Status | Tags |
|---|---|---|---|---|
gpt-4o | 128K | ¥30/M tokens | Available | MultimodalStrong Reasoning |
gpt-4o-mini | 128K | ¥3/M tokens | Available | Cost-Effective |
gpt-4-turbo | 128K | ¥25/M tokens | Available | Stable |
gpt-3.5-turbo | 16K | ¥1/M tokens | Available | Classic |
Anthropic Claude
Long-text processing, creative writing, safety-sensitive scenarios
| Model | Context Length | Pricing | Status | Tags |
|---|---|---|---|---|
claude-3-5-sonnet | 200K | ¥15/M tokens | Available | BalancedLatest |
claude-3-opus | 200K | ¥50/M tokens | Available | Top Reasoning |
claude-3-sonnet | 200K | ¥15/M tokens | Available | Balanced |
claude-3-haiku | 200K | ¥3/M tokens | Available | Fast |
Image Generation
Artistic-style creation, precise instruction following
| Model | Context Length | Pricing | Status | Tags |
|---|---|---|---|---|
midjourney | — | ¥0.5/张 | Available | ImageArtistic |
dall-e-3 | — | ¥0.3/张 | Available | Image |
How to Choose a Model
Quickly narrow it down from two angles: task type and budget.
By Task
| Scenario | Recommended Model |
|---|---|
| Everyday chat | deepseek-chat (best value) |
| Code generation | deepseek-reasoner / gpt-4o |
| Math reasoning | deepseek-reasoner / claude-3-opus |
| Long-document analysis | gemini-3-pro-preview (1M context) |
| Creative writing | claude-3-5-sonnet |
| English tasks | gpt-4o / claude-3-5-sonnet |
| Chinese tasks | deepseek-chat |
| Image understanding | gemini-3-pro-image-preview / gpt-4o |
| Image generation | dall-e-3 / midjourney |
By Budget
| Budget Range | Recommended Model |
|---|---|
| Low (< ¥5/M tokens) | deepseek-chat, gpt-3.5-turbo, gemini-3-flash-preview |
| Medium (¥5–20/M tokens) | claude-3-sonnet, gemini-3-pro-preview, gpt-4o-mini |
| High (> ¥20/M tokens) | gpt-4o, claude-3-opus |
Switching Models
All models share the same API format. Just change the model field (the rest of your code stays the same): client.chat.completions.create(model="deepseek-chat", ...) → model="gemini-3-pro-preview".
Things to note
- Model availability: A
Betastatus means a preview release whose interface may change. - Quota limits: Different API keys may have different model access permissions.
- Price changes: Pricing changes as providers adjust it; the console display is authoritative.
- Context limits: Requests that exceed the context length are truncated or return an error.