Cogito Models by Deep Cogito
Last refreshed 2026-06-29. Next refresh: weekly.
Details
Capabilities
About
The Cogito family is a series of hybrid open-weight reasoning models from Deep Cogito, trained with Iterated Distillation and Amplification (IDA). Models span 3B to 671B parameters, support both direct and extended-thinking (reasoning) modes, and are fine-tuned from Llama and Qwen base checkpoints (v1 Preview) and DeepSeek V3 Base (v2.1). Available via Fireworks AI, Together AI, and other inference providers.
Decision facts
- Best fit
- reasoningJSON / Tool usestructured outputs
- Capability starting point
- Cogito v2.1 671B with 128k context and reasoning, JSON / Tool use, and structured outputs
- Lowest tracked input
- Cogito v1 Preview Llama 3B · $0.100/1M · Fireworks AI
- Closest related family
- 360Zhinao
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs 128k context, 671B parameters, and reasoning.
Use when the workload needs 671B parameters, reasoning, and JSON / Tool use.
Use when the workload needs 109B parameters, reasoning, and JSON / Tool use.
Use when the workload needs 405B parameters, reasoning, and JSON / Tool use.
Use when the workload needs 70B parameters, reasoning, and JSON / Tool use.
Use when the workload needs 128k context, 3B parameters, and reasoning.
Use when the workload needs 128k context, 70B parameters, and reasoning.
Use when the workload needs 128k context, 8B parameters, and reasoning.
Use when the workload needs 128k context, 14B parameters, and reasoning.
Use when the workload needs 128k context, 32B parameters, and reasoning.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Cogito v2.1 671B | Use when the workload needs 128k context, 671B parameters, and reasoning. | 2025-11 | 128k context671B parametersreasoning | Current |
| Cogito v2 Preview DeepSeek 671B MoE | Use when the workload needs 671B parameters, reasoning, and JSON / Tool use. | 2025-07 | 671B parametersreasoningJSON / Tool use | Current |
| Cogito v2 Preview Llama 109B MoE | Use when the workload needs 109B parameters, reasoning, and JSON / Tool use. | 2025-07 | 109B parametersreasoningJSON / Tool use | Current |
| Cogito v2 Preview Llama 405B | Use when the workload needs 405B parameters, reasoning, and JSON / Tool use. | 2025-07 | 405B parametersreasoningJSON / Tool use | Current |
| Cogito v2 Preview Llama 70B | Use when the workload needs 70B parameters, reasoning, and JSON / Tool use. | 2025-07 | 70B parametersreasoningJSON / Tool use | Current |
| Cogito v1 Preview Llama 3B | Use when the workload needs 128k context, 3B parameters, and reasoning. | 2025-04 | 128k context3B parametersreasoning | Current |
| Cogito v1 Preview Llama 70B | Use when the workload needs 128k context, 70B parameters, and reasoning. | 2025-04 | 128k context70B parametersreasoning | Current |
| Cogito v1 Preview Llama 8B | Use when the workload needs 128k context, 8B parameters, and reasoning. | 2025-04 | 128k context8B parametersreasoning | Current |
| Cogito v1 Preview Qwen-14B | Use when the workload needs 128k context, 14B parameters, and reasoning. | 2025-04 | 128k context14B parametersreasoning | Current |
| Cogito v1 Preview Qwen-32B | Use when the workload needs 128k context, 32B parameters, and reasoning. | 2025-04 | 128k context32B parametersreasoning | Current |
Release Timeline
3 release groupsSpecifications(10 models)
| Model | Released | Context | Parameters | Reasoning | JSON / Tool use | Structured Outputs | Code Exec |
|---|---|---|---|---|---|---|---|
| Cogito v2.1 671B | 2025-11 | 128k | 671B | Yes | Yes | Yes | Yes |
| Cogito v2 Preview DeepSeek 671B MoE | 2025-07 | — | 671B | Yes | Yes | No | No |
| Cogito v2 Preview Llama 109B MoE | 2025-07 | — | 109B | Yes | Yes | No | No |
| Cogito v2 Preview Llama 405B | 2025-07 | — | 405B | Yes | Yes | No | No |
| Cogito v2 Preview Llama 70B | 2025-07 | — | 70B | Yes | Yes | No | No |
| Cogito v1 Preview Llama 3B | 2025-04 | 128k | 3B | Yes | Yes | Yes | No |
| Cogito v1 Preview Llama 70B | 2025-04 | 128k | 70B | Yes | Yes | Yes | No |
| Cogito v1 Preview Llama 8B | 2025-04 | 128k | 8B | Yes | Yes | Yes | No |
| Cogito v1 Preview Qwen-14B | 2025-04 | 128k | 14B | Yes | Yes | Yes | No |
| Cogito v1 Preview Qwen-32B | 2025-04 | 128k | 32B | Yes | Yes | Yes | No |
Available From(1 provider)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Cogito v1 Preview Llama 3B | Fireworks AI | $0.1 | $0.1 | Serverless |
| Cogito v1 Preview Llama 8B | Fireworks AI | $0.2 | $0.2 | Serverless |
| Cogito v1 Preview Qwen-14B | Fireworks AI | $0.2 | $0.2 | Serverless |
| Cogito v1 Preview Llama 70B | Fireworks AI | $0.9 | $0.9 | Serverless |
| Cogito v1 Preview Qwen-32B | Fireworks AI | $0.9 | $0.9 | Serverless |
Models(10)
Cogito v2.1 671B
Cogito v2 Preview DeepSeek 671B MoE
Cogito v2 Preview Llama 109B MoE
Cogito v2 Preview Llama 405B
Cogito v2 Preview Llama 70B
Cogito v1 Preview Llama 3B
Cogito v1 Preview Llama 70B
Cogito v1 Preview Llama 8B
Cogito v1 Preview Qwen-14B
Cogito v1 Preview Qwen-32B