<!-- LLM_VERSION_INFO
FORMAT: text/markdown
CONTENT_TYPE: article
ORIGINAL_URL: https://www.vals.ai/models/arcee-ai_trinity-large-thinking
ALTERNATE_VERSION: models/arcee-ai_trinity-large-thinking/index.html (text/html)
EXTRACTION_DATE: 2026-04-17T00:42:54.428Z

This is the markdown version with text-only content (images converted to alt-text).
For rich formatting with images, request the HTML version at: models/arcee-ai_trinity-large-thinking/index.html
-->

# Open Weights & Proprietary

## Models

| Release Date | Model Name                         | Company      | Image                                        |
|--------------|-----------------------------------|--------------|---------------------------------------------|
| 4/16/2026    | Claude Opus 4.7                   | Anthropic    |  |
| 4/8/2026     | Muse Spark                        | Meta         |          |
| 4/2/2026     | Gemma 4 31B IT                   | Google       |      |
| 4/2/2026     | Qwen 3.6 Plus                    | Alibaba      |    |
| 4/1/2026     | GLM 5.1                          | zAI          |            |
| 4/1/2026     | Trinity Large Thinking            | Arcee AI     |  |
| 3/17/2026    | GPT 5.4 Mini                     | OpenAI       |      |
| 3/17/2026    | GPT 5.4 Nano                     | OpenAI       |      |
| 3/17/2026    | MiniMax-M2.7                     | MiniMax      |    |
| 3/9/2026     | Grok 4.20 (Reasoning)            | xAI          |            |
| 3/5/2026     | GPT 5.4                          | OpenAI       |      |
| 3/3/2026     | Gemini 3.1 Flash Lite Preview    | Google       |      |
| 2/24/2026    | GPT 5.3 Codex                    | OpenAI       |      |
| 2/23/2026    | Qwen 3.5 Flash                   | Alibaba      |    |
| 2/19/2026    | Gemini 3.1 Pro Preview (02/26)   | Google       |      |
| 2/17/2026    | Claude Sonnet 4.6                | Anthropic    |  |
| 2/16/2026    | Qwen 3.5 Plus                    | Alibaba      |    |
| 2/12/2026    | MiniMax-M2.5                     | MiniMax      |    |
| 2/11/2026    | MiniMax-M2.5                     | MiniMax      |    |
| 2/5/2026     | GLM 5                            | zAI          |            |
| 2/5/2026     | Claude Opus 4.6 (Nonthinking)    | Anthropic    |  |
| 2/5/2026     | Claude Opus 4.6 (Thinking)       | Anthropic    |  |
| 1/26/2026    | Kimi K2.5                        | Moonshot AI  |  |
| 1/23/2026    | Qwen 3 Max Thinking               | Alibaba      |    |
| 12/23/2025   | MiniMax-M2.1                     | MiniMax      |    |
| 12/22/2025   | GLM 4.7                          | zAI          |            |
| 12/17/2025   | Gemini 3 Flash (12/25)           | Google       |      |
| 12/17/2025   | MiMo V2 Flash                    | Xiaomi       |      |
| 12/11/2025   | GPT 5.2                          | OpenAI       |      |
| 12/11/2025   | GPT 5.2 Codex                    | OpenAI       |      |

### Trinity Large Thinking

- Release Date: 4/1/2026
- Default Provider: Arcee-Ai

#### Benchmarks

| Metric                    | Value        |
|---------------------------|--------------|
| Accuracy (Vals Index)     | 42.60% ± 1.88 |
| Latency (Vals Index)      | 380.48s      |
| Cost/Test (Vals Index)    | $0.14        |
| Context Window             | 262k         |
| Max Output Tokens          | 80k          |

- Temperature: 0.3
- Top P: 0.95
- Top K: Default
- Max Output Tokens: 80,000

#### Accuracy Rankings

| Benchmark                  | Accuracy        |
|---------------------------|------------------|
| Vals Index                | -53.33% ± 1.88  |
| CaseLaw (v2)             | -88.40% ± 1.86  |
| CorpFin                   | -103.42% ± 0.98 |
| Terminal-Bench 2.0       | -41.17% ± 4.09  |

### Proprietary Benchmarks

Academic Benchmarks: Read about our [methodology](/content/methodology/index.html).