Best DeepSeek Distill for Your GPU (2026)
Match your GPU to a DeepSeek-R1 distill in one line: RTX 3060 12GB runs the 7B, RTX 4060 Ti 16GB runs the 14B, RTX 4090 runs the 32B (beats o1-mini in DeepSeek's published results), and dual-GPU runs the 70B. On an 8 GB card, run DeepSeek-R1-0528-Qwen3-8B — among the strongest small reasoning distills at its release. All commands default to Q4_K_M. Status: R1 is no longer DeepSeek's current generation (DeepSeek-V4 is), but these distills are separate open-weight releases that still run locally. The o1-mini comparison is historical: OpenAI shut o1-mini down on October 27, 2025.
Key Takeaways
- ✓R1 is no longer DeepSeek's current generation — DeepSeek-V4 (Flash and Pro, MIT-licensed open weights) is. The R1 distills are separate open-weight releases and still run locally.
- ✓RTX 3060 12GB → 7B distill; RTX 4060 Ti 16GB → 14B (the sweet spot); RTX 4090 → 32B (beats o1-mini).
- ✓Dual-GPU or 48 GB → 70B distill, the strongest of the six.
- ✓On 8 GB, the best small model is DeepSeek-R1-0528-Qwen3-8B.
- ✓Every model installs at Q4_K_M with one command, e.g. `ollama run deepseek-r1:14b`.
- ✓Set temperature to 0.6 and use no system prompt, as DeepSeek recommends, to avoid endless repetition.
- ✓This is the R1 reasoning family — not DeepSeek-V3, which is a chat model.
This page contains links to third-party products for reference. PromptQuorum is not enrolled in any affiliate program — these are plain links that earn no commission. Clicking links and your next steps are entirely your own responsibility. These links do not represent any endorsement or verification by PromptQuorum.
Quick Answer
Find your card: RTX 3060 12GB → 7B, RTX 4070 12GB → 8B (14B only with short context), RTX 4060 Ti 16GB / 4070 Ti Super / 4080 → 14B, RTX 4090 → 32B, dual-GPU/48 GB → 70B. The 32B needs ~20 GB for the weights alone, so it does not fit 12 or 16 GB cards. For the best small model on 8 GB, run DeepSeek-R1-0528-Qwen3-8B (`ollama run deepseek-r1:8b`). Each runs with one Ollama command at Q4_K_M.
- ▸RTX 3060 12GB → deepseek-r1:7b — ~30–40 tok/s
- ▸RTX 4060 Ti 16GB → deepseek-r1:14b — ~25–35 tok/s (recommended)
- ▸RTX 4070 12GB → deepseek-r1:8b (14B only at short context) — ~30–40 tok/s
- ▸RTX 4070 Ti Super / 4080 16GB → deepseek-r1:14b — ~40–50 tok/s (32B does not fit)
- ▸RTX 4090 24GB → deepseek-r1:32b — ~30–40 tok/s, beats o1-mini
- ▸Dual-GPU / 48 GB → deepseek-r1:70b — ~12–18 tok/s
- ▸8 GB card, best small → DeepSeek-R1-0528-Qwen3-8B
GPU → DeepSeek-R1 Distill → Ollama Command
Find the GPU you own in the first column and read across. The tok/s figures are rough indicative estimates for Q4_K_M reasoning workloads, not PromptQuorum measurements, and vary with context length, sampling settings and hardware. When two models fit, the larger one reasons better; the smaller one is faster.
| GPU (VRAM) | Best Distill | Ollama Command | Expected tok/s |
|---|---|---|---|
| RTX 3060 12GB (8 GB tier) | DeepSeek-R1-Distill-Qwen-7B | ollama run deepseek-r1:7b | ~30–40 |
| 8 GB, best small | DeepSeek-R1-0528-Qwen3-8B | ollama run deepseek-r1:8b | ~30–40 |
| RTX 4060 Ti 16GB | DeepSeek-R1-Distill-Qwen-14B | ollama run deepseek-r1:14b | ~25–35 |
| RTX 4070 12GB | 8B, or 14B (short context) | ollama run deepseek-r1:8b | ~30–40 |
| RTX 4070 Ti Super / 4080 16GB | 14B (32B does not fit) | ollama run deepseek-r1:14b | ~40–50 |
| RTX 4090 24GB | DeepSeek-R1-Distill-Qwen-32B | ollama run deepseek-r1:32b | ~30–40 |
| Dual-GPU / 48 GB | DeepSeek-R1-Distill-Llama-70B | ollama run deepseek-r1:70b | ~12–18 |
How to Use This Table in 3 Steps
Four lines: (1) find your GPU and its VRAM, (2) install Ollama and run the matching command, (3) confirm with `ollama ps` that the model sits fully on the GPU, (4) set temperature 0.6 and clear the system prompt. If a model is too slow, drop one tier; if you have spare VRAM, move up a tier for better reasoning.
V3 vs R1: This Table Is R1 Only
**DeepSeek-R1 is the reasoning family these commands install; DeepSeek-V3 is a separate chat model.** Do not expect a V3 experience from these distills — they are tuned to show step-by-step reasoning for math and logic. V3 is also a 671B MoE and not consumer-runnable; see the [DeepSeek-V3 hardware bite](/prompt-bites/deepseek-v3-local-hardware-requirements).
**Update:** DeepSeek's current generation is DeepSeek-V4 (Flash and Pro), released as open weights under the MIT license (preview: April 24, 2026). DeepSeek's API now serves V4: the deepseek-chat (V3-generation) and deepseek-reasoner (R1-generation) aliases were announced as fully retired after July 24, 2026. The R1 distills are separate open-weight releases that were not part of that notice, and they still run locally. V4 itself is not a consumer-GPU model: V4-Pro has 1.6T total parameters (49B active) and the original V4-Flash 284B (13B active), so even at 4-bit precision (parameter count × 0.5 byte) the weights alone need roughly 140 GB (Flash) to 800 GB (Pro), and DeepSeek's model cards list no GPU requirement. For reasoning on a single consumer GPU, the distills remain the practical pick.
Related Guides
- ▸DeepSeek-R1 Distill VRAM Cheatsheet — every distill by quant (Q4_K_M, Q8, FP16) with VRAM and min-GPU
- ▸Best Local Reasoning Model 2026: DeepSeek-R1 Ranked — the full ranked guide with benchmarks and tiers
- ▸DeepSeek-V3 Local Hardware Requirements — the V3 chat-model counterpart
Frequently Asked Questions
Is DeepSeek-R1 still DeepSeek's current model?▾
What DeepSeek distill runs on an RTX 4090?▾
Which DeepSeek-R1 distill fits 16 GB of VRAM?▾
Can an RTX 4070 run deepseek-r1:32b?▾
What is the best DeepSeek distill for an 8 GB GPU?▾
Why is my distill slow?▾
Do I need to pick a quantization?▾
Want the full breakdown?
Read the complete guide →Related Prompt Bites