Trinity Large Thinking
Summary
Trinity Large Thinking is an open-weight language model from Arcee AI, released on 1 Apr 2026. It accepts text and generates text, with a 262K-token context window and up to 80K output tokens per response. It is scheduled for retirement on 26 Oct 2026.
Reasoning-optimized 398B MoE agent model with extended thinking for long-horizon and multi-turn tool use
Last verified 18 Nov 2026. Every figure on this page links to its source.
Specifications
| Developer | Arcee AI |
|---|---|
| Release date | 1 Apr 2026 [models.dev] |
| Status | Preview [models.dev] |
| Weights | Open weights [Hugging Face] |
| License | openmdw-1.1 [Hugging Face] |
| Parameters | 399B [Hugging Face] |
| Context window | 262K (262,144 tokens) [models.dev] |
| Max output | 80K tokens [models.dev] |
| Input | Text [models.dev] |
| Output | Text [models.dev] |
| Reasoning mode | Yes [models.dev] |
| Tool calling | Yes [models.dev] |
| Structured output | No [models.dev] |
| Retirement date | 26 Oct 2026 [OpenRouter] |
| Hugging Face | arcee-ai/Trinity-Large-Thinking [Hugging Face] |
Trinity Large Thinking API pricing by provider
USD per million tokens. Sorted by input price.
We do not currently track a paid API listing for this model.
Running Trinity Large Thinking locally
Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.
| Quantization | Weights | Total, 8K context | Total, 32K context |
|---|---|---|---|
| Q4_K_M (4-bit) | 242 GB | 244 GB | 245 GB |
| Q8_0 (8-bit) | 424 GB | 426 GB | 427 GB |
| FP16 / BF16 (16-bit) | 797 GB | 799 GB | 801 GB |
At 4-bit with an 8K conversation, Trinity Large Thinking needs about 244 GB, more than any single consumer graphics card; it needs a unified-memory machine, a multi-GPU setup or a split between graphics card and system RAM. Check your own hardware, or see Best computer for local AI on our sister site AIPCs.
Frequently asked questions
What is the context window of Trinity Large Thinking?
Trinity Large Thinking has a context window of 262,144 tokens (262K) and can generate up to 80,000 tokens in a single response, according to models.dev.
Is Trinity Large Thinking open source?
Trinity Large Thinking is an open-weight model: its weights are published on Hugging Face as arcee-ai/Trinity-Large-Thinking under the openmdw-1.1 license. Check the license terms before commercial use.
When was Trinity Large Thinking released?
Arcee AI released Trinity Large Thinking on 1 Apr 2026, according to models.dev.
How much VRAM does Trinity Large Thinking need?
About 244 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 426 GB at 8-bit. That is more than any single consumer graphics card holds. These are estimates from the model's published parameter count and architecture.
Similar models from other labs
Comparable on output type, weights, release date, price and context window. Prices are the lowest listed, USD per million tokens.
| Model | Lab | Released | Context | Input $/1M | Output $/1M | Weights |
|---|---|---|---|---|---|---|
| Gemma 4 26B A4B IT | Google DeepMind | 2 Apr 2026 | 262K | $0.07 | $0.30 | Open |
| Gemma 4 31B IT | Google DeepMind | 2 Apr 2026 | 262K | $0.14 | $0.40 | Open |
| Step 3.5 Flash 2603 | StepFun | 2 Apr 2026 | 256K | – | – | Open |
| Mistral Small 4 | Mistral AI | 16 Mar 2026 | 262K | $0.15 | $0.60 | Open |
| Qwen3.6 35B-A3B | Alibaba (Qwen) | 17 Apr 2026 | 262K | $0.10 | $0.90 | Open |
| Hy3 preview | Tencent (Hunyuan) | 20 Apr 2026 | 262K | $0.18 | $0.60 | Open |
Change history
- Context window changed from 128K to 262K
- Name changed from Arcee Trinity Large Thinking (Public Preview) to Trinity Large Thinking
- Status changed from active to preview
- Max output changed from 32K to 80K
- Name changed from Trinity Large Thinking to Arcee Trinity Large Thinking (Public Preview)
- Status changed from preview to active
- Max output changed from 80K to 32K
- Context window changed from 262K to 128K
- Max output changed from 32K to 80K
- Status changed from active to preview
- Context window changed from 128K to 262K
- Name changed from Arcee Trinity Large Thinking (Public Preview) to Trinity Large Thinking
- Name changed from Trinity Large Thinking to Arcee Trinity Large Thinking (Public Preview)
- Status changed from preview to active
- Context window changed from 262K to 128K
- Max output changed from 80K to 32K
- Max output changed from 32K to 80K
- Name changed from Arcee Trinity Large Thinking (Public Preview) to Trinity Large Thinking
- Status changed from active to preview
- Context window changed from 128K to 262K
- Status changed from preview to active
- Context window changed from 262K to 128K
- Name changed from Trinity Large Thinking to Arcee Trinity Large Thinking (Public Preview)
- Max output changed from 80K to 32K
- Name changed from Arcee Trinity Large Thinking (Public Preview) to Trinity Large Thinking
Sources
- https://openrouter.ai/arcee-ai/trinity-large-thinking
- https://github.com/sst/models.dev/blob/dev/providers/arcee/models/trinity-large-thinking.toml
- https://huggingface.co/arcee-ai/Trinity-Large-Thinking
- https://github.com/sst/models.dev/blob/dev/providers/nano-gpt/models/arcee-ai/trinity-large-thinking.toml