Vendor · Language models
DeepReinforce has 5 models in the directory: 5 language models.
In the directory: 5 language models. 5 of them with open weights that can run locally.
Every model there ever was. Models with a profile lead to every benchmark, price and source.
| Ornith 1.5 35B A3B | Aug 18, 2026 | Active | n/a | n/a | n/a | 262K | 42 |
| Ornith 1.0 31B | Jun 25, 2026 | Active | n/a | n/a | n/a | 262K | 25 |
| Ornith 1.0 35B | Jun 25, 2026 | Active | n/a | n/a | n/a | 262K | 55 |
| Ornith 1.0 397B | Jun 25, 2026 | Active | n/a | n/a | n/a | 262K | 46 |
| Ornith 1.0 9B | Jun 25, 2026 | Active | n/a | n/a | n/a | 262K | 42 |
Language model · Aug 18, 2026
New in Ornith 1.5 35B A3B is the expanded self-improvement loop described by its model card. Tasks, agent scaffolds, and solution rollouts are jointly generated through reinforcement learning; 35 billion total parameters sit against about 3 billion active. A 262K context, reasoning, tool calling, and OpenAI-compatible serving target terminal coding while A3B limits active compute.
Language model · Jun 25, 2026
Ornith 1.0 31B is recorded on the official family card as a 31B-dense variant. The documented distinction of the line is a reinforcement-learning method that optimizes both solution rollouts and the agent scaffolds used to produce them. No 31B-specific card provides its own context, modality, or tool-calling details.
Language model · Jun 25, 2026
Ornith 1.0 35B combines a sparse MoE with a 262K context plus image and text input. Only part of its 35 billion total parameters is active per token, which is why the model card positions it as a comparatively light single-GPU option for coding agents. The card also names terminal agents.
Language model · Jun 25, 2026
Ornith 1.0 397B is the large MoE tier in the first Ornith family, with 397 billion total parameters. The checkpoint accepts image and text input with a 262K context. The defining training idea remains joint reinforcement-learning improvement of solution paths and agent scaffolds.
Language model · Jun 25, 2026
Ornith 1.0 9B is the family's smallest dense member and accepts image and text input with a 262K-token context. The model card explicitly targets terminal-based coding agents and OpenAI-compatible local servers.
| Vendor | Models | Model types | Best language model | Top 10 | Open weights | New in 2026 |
|---|---|---|---|---|---|---|
| Anthropic | 33 | 33 language models | Claude Opus 5.5 (Rank 1 to 3) | 4 | 0 | 13 |
| OpenAI | 63 | 46 language models, 4 image models, 2 video models, 9 audio models, 2 specialist models | GPT-6 Astra (Rank 4 to 8) | 2 | 4 | 27 |
| 66 | 43 language models, 4 image models, 6 video models, 11 audio models, 2 specialist models | Gemini 4 Carbon (Rank 1 to 5) | 2 | 19 | 26 | |
| Meta | 32 | 29 language models, 1 image model, 1 video model, 1 specialist model | Muse Wonder Full (Rank 6 to 9) | 2 | 26 | 7 |
| DeepSeek | 23 | 23 language models | DeepSeek V4.1 Flash 1017 (Rank 13 to 20) | 0 | 23 | 9 |
| Moonshot AI | 12 | 12 language models | Kimi K3.1 Pro (Rank 10 to 13) | 0 | 10 | 7 |
| xAI | 16 | 12 language models, 2 image models, 2 video models | Grok 4.7 (Rank 18 to 24) | 0 | 1 | 12 |
| DeepReinforce | 5 | 5 language models | not ranked yet | 0 | 5 | 5 |
By release date. Grey: retired models.
5 releases since January 2026, by release date.
The directory lists 5 Ornith models from DeepReinforce: 5 language models. It counts every model DeepReinforce ever released, including retired ones. 5 are currently available, the newest being Ornith 1.5 35B A3B, Ornith 1.0 31B and Ornith 1.0 35B.
Ornith 1.5 35B A3B, released on Aug 18, 2026. The timeline below shows every release by date.
Yes, 5 of the 5 models have open weights. Each model profile lists how much memory it needs.
Up to 262,144 tokens (≈ 390 A4 pages), which 5 of the current models reach, including Ornith 1.5 35B A3B. Every model's value is in the model list.
Data as of Nov 1, 2026