Cogito v1 70B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
Largest Cogito variant for multi-GPU workstations. Scores 65.0% on AIME 2025 — comparable to DeepSeek-R1 671B on math benchmarks. Requires ~40 GB at Q4 (dual RTX 3090 or A6000). Best open-source reasoning model for teams with workstation hardware.
Cogito v1 70B needs about 43 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Includes the KV cache at 8K context, so it reads higher than the headline figure.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 26.5 GB | ~4 tok/s | Offload | |
| Q3_K_M 3.41 bpw | 33.3 GB | ~3 tok/s | Offload | |
| Q4_K_M 4.83 bpw | 45.7 GB | ~3 tok/s | Offload | |
| Q5_K_M 5.67 bpw | 53.1 GB | ~2 tok/s | Offload | |
| Q6_K 6.56 bpw | 60.9 GB | — | Too big | |
| Q8_0 8.50 bpw | 77.9 GB | — | Too big | |
| F16 16.00 bpw | 143.5 GB | — | Too big |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Cogito v1 70B VRAM calculator →
Get Cogito v1 70B running
The cheapest catalogued GPU that runs Cogito v1 70B is the NVIDIA DGX Spark (128 GB).
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run Cogito v1 70B
Install Ollama, then run:
ollama run cogito:70bSpecifications
Curated — A hand-written entry from before this catalogue recorded its sources. The figures are long-standing but their provenance is not on file.
- Parameters
- 70 Billion
- Context window
- 128,000
- Architecture
- Hybrid Reasoning Transformer
- Provider
- Deep Cogito
- Licence
- Apache 2.0
- Practical deployment
- Workstation (large unified memory or pro GPU)
- Specified at
- Q4_K_M
- System RAM
- 64 GB
- Record updated
- 2026-03-20
Commercial use permitted. No usage restrictions beyond attribution.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
| Benchmark | Score | Provenance |
|---|---|---|
| MMLU | 88.9 / 100 % | reported |
| MATH | 83.7 / 100 % | reported |
| AIME 2025 | 65 / 100 % | reported |
| HumanEval | 88.4 / 100 % | reported |
Can I run Cogito v1 70B on my GPU?
- Cogito v1 70B on AMD Radeon RX 7800 XT
- Cogito v1 70B on AMD Radeon RX 7900 XT
- Cogito v1 70B on AMD Radeon RX 7900 XTX
- Cogito v1 70B on AMD Radeon RX 9060 XT 16GB
- Cogito v1 70B on AMD Radeon RX 9060 XT 8GB
- Cogito v1 70B on AMD Radeon RX 9070
- Cogito v1 70B on AMD Radeon RX 9070 XT
- Cogito v1 70B on AMD Ryzen AI Max+ 395
- Cogito v1 70B on Apple M1
- Cogito v1 70B on Apple M1 Max
- Cogito v1 70B on Apple M1 Pro
- Cogito v1 70B on Apple M2 Max
- Cogito v1 70B on Apple M2 Pro
- Cogito v1 70B on Apple M3 Max
- Cogito v1 70B on Apple M3 Pro
- Cogito v1 70B on Apple M4
- Cogito v1 70B on Apple M5
- Cogito v1 70B on Apple M5 Pro
- Cogito v1 70B on Intel Arc B570
- Cogito v1 70B on Intel Arc B580
- Cogito v1 70B on NVIDIA A100 80GB (PCIe)
- Cogito v1 70B on NVIDIA GeForce RTX 3060 (12GB)
- Cogito v1 70B on NVIDIA GeForce RTX 3070
- Cogito v1 70B on NVIDIA GeForce RTX 3070 Ti
- Cogito v1 70B on NVIDIA GeForce RTX 3080 (10GB)
- Cogito v1 70B on NVIDIA GeForce RTX 3090
- Cogito v1 70B on NVIDIA GeForce RTX 4060
- Cogito v1 70B on NVIDIA GeForce RTX 4060 Ti 16GB
- Cogito v1 70B on NVIDIA GeForce RTX 4070
- Cogito v1 70B on NVIDIA GeForce RTX 4070 Super
- Cogito v1 70B on NVIDIA GeForce RTX 4070 Ti
- Cogito v1 70B on NVIDIA GeForce RTX 4070 Ti Super
- Cogito v1 70B on NVIDIA GeForce RTX 4080
- Cogito v1 70B on NVIDIA GeForce RTX 4080 Super
- Cogito v1 70B on NVIDIA GeForce RTX 4090
- Cogito v1 70B on NVIDIA GeForce RTX 5060
- Cogito v1 70B on NVIDIA GeForce RTX 5060 Ti 16GB
- Cogito v1 70B on NVIDIA GeForce RTX 5060 Ti 8GB
- Cogito v1 70B on NVIDIA GeForce RTX 5070
- Cogito v1 70B on NVIDIA GeForce RTX 5070 Ti
- Cogito v1 70B on NVIDIA GeForce RTX 5080
- Cogito v1 70B on NVIDIA GeForce RTX 5090
- Cogito v1 70B on NVIDIA H100 80GB (PCIe)
- Cogito v1 70B on NVIDIA L40S
- Cogito v1 70B on NVIDIA RTX 6000 Ada Generation
- Cogito v1 70B on NVIDIA RTX PRO 6000 Blackwell
Other Cogito v1 sizes
Cogito v1 70B — frequently asked questions
How much VRAM does Cogito v1 70B need?
About 43 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does Cogito v1 70B run on an RTX 4090 (24 GB)?
No. Cogito v1 70B needs about 43 GB at Q4_K_M, more than a single RTX 4090's 24 GB. It needs a larger card, several GPUs, or Apple Silicon with enough unified memory — or it runs with part of the weights offloaded to system RAM, which is much slower.
How do I run Cogito v1 70B locally?
Install Ollama and run `ollama run cogito:70b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.
What other sizes does Cogito v1 come in?
Cogito v1 3B (3 GB), Cogito v1 8B (6 GB), Cogito v1 14B (9 GB), Cogito v1 32B (20 GB), Cogito v1 70B (43 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.