Multiple AI Models
Every Tridan AI assistant can run on models from four providers — OpenAI, Anthropic, Meta, and DeepSeek. You pick the model per assistant, so a simple FAQ bot and a research agent can each use the model that fits.
Model providers
OpenAI
GPT-5.5, GPT-5.4, GPT-5.5 MiniThe current GPT-5.5 family. Strong general-purpose reasoning and reliable tool calling, with Mini and Nano variants for high-volume workloads.
Anthropic
Claude Opus 4.8, Claude Sonnet 4.6, Claude Haiku 4.5Excellent long-context comprehension and careful, well-structured answers. Opus for the hardest tasks, Sonnet for everyday complex work, Haiku for speed.
Meta
Llama 4 Maverick, Llama 3.3 70BOpen-weight models served through Groq fast inference, which makes them the lowest-latency option on the platform.
DeepSeek
DeepSeek V3Strong analytical and reasoning performance at a low price point.
Model tiers
Models are grouped into two tiers. Which tiers you can use depends on your plan (see the table below).
Basic Models
Cost-effective models for general tasks
- GPT-5.5 Mini
- Claude Haiku 4.5
- Llama 3.3 70B
- DeepSeek V3
Premium Models
Advanced models for complex tasks
- GPT-5.5
- GPT-5.4
- Claude Opus 4.8
- Claude Sonnet 4.6
- Llama 4 Maverick
- DeepSeek V3
Choosing a model
Start with a Basic model and move up only where quality demands it — each assistant can use a different model, so you only pay premium rates where they matter.
Swipe sideways to see the full table
| Task | Recommended model | Why |
|---|---|---|
| FAQ and high-volume support | GPT-5.5 Mini or Claude Haiku 4.5 | Fast and inexpensive; quality is more than enough for grounded Q&A over your knowledge base. |
| Complex reasoning and analysis | Claude Opus 4.8 or GPT-5.5 | Best multi-step reasoning for research, planning, and difficult edge cases. |
| Long documents and nuanced writing | Claude Sonnet 4.6 | Strong long-context comprehension with a good balance of quality, speed, and cost. |
| Lowest response latency | Llama 3.3 70B (via Groq) | Groq inference returns tokens fastest, which matters for real-time chat and voice. |
| Cost-sensitive advanced tasks | DeepSeek V3 | Premium-tier reasoning at a fraction of the usual cost. |
| Agents that call tools | GPT-5.5 or Claude Sonnet 4.6 | Most dependable structured output and tool-call accuracy. |
Model access by plan
Swipe sideways to see the full table
| Plan | Basic models | Premium models |
|---|---|---|
| Free Plan | Included | — |
| Starter Plan | Included | — |
| Business Plan | Included | Included |
| Enterprise Plan | Included | Included |
See Plan Consumption & AI Models for token limits and how usage is metered.
Tips
- Test with your own prompts. Run the same question through two or three models in the assistant playground before committing.
- Mix models per assistant. A support bot on GPT-5.5 Mini and a contract-analysis agent on Claude Opus 4.8 can share one workspace.
- Premium models consume tokens faster. Reserve them for assistants where answer quality is business-critical.
Premium models are available on the Business Plan and Enterprise Plan. Compare limits and pricing across all plans.
View all plans