DeepSeek V4 Flash 284B: the 1st frontier model that fits on a Mac Studio

DeepSeek V4 Flash 284B was released on April 24, 2026, at the same time as V4 Pro. It is the "efficient" variant: 284 billion total parameters in MoE (13B active per token), an MIT license, a 1M-token context, the same CSA+HCA architecture, and the same 3 thinking modes as Pro. Most importantly, it is the first frontier-class model that fits on a single workstation: 170 GB in Q4—a 256 GB Mac Studio M3 Ultra is enough. This guide explains how to install it and what it is worth.
#DeepSeek V4 Flash at a glance
- Output
- April 24, 2026, simultaneous with V4 Pro. Available on HuggingFace: deepseek-ai/DeepSeek-V4-Flash.
- Architecture
- MoE: 284B total, 13B active per token. 128 experts per layer, top-8 routing. Base + Instruct variants available.
- Innovations inherited from the Pro
- Caution: CSA+HCA, mHC, Muon optimizer, mixed FP4+FP8 training.
- Context
- 1 000 000 native tokens (same as Pro).
- Thinking modes
- 3 levels: No / High / Max — toggleable in the prompt.
- License
- MIT, same as Pro. No commercial or geographic usage restrictions.
#1. Why it’s an event
Before April 24, 2026, running a frontier model locally meant either a GPU cluster costing €80k+ or accepting the degraded quality of a 70B model. DeepSeek V4 Flash changes that: 170 GB in Q4 fits configurations costing €4k–8k used.
Benefit from our experience and a guided path to build your local assistant and use your AI every day.
- Lifetime online access
- PDF + files
- Refunded within 30 days
- Mac Studio M3 Ultra 256 GB
- €7,299 listed through September 2026, now available secondhand (the M5 Ultra 256 GB replaces it at Apple). 170 GB model + 30 GB context = plenty of room. ~10–12 tok/s in Q4.
- 2× RTX 5090 32 GB (64 GB VRAM)
- ≈ €11,400. Insufficient—~110 GB is missing. Possible with CPU offload, but slow (3-5 tok/s).
- 4× RTX A6000 48 GB (192 GB VRAM)
- ~€12,000 used. Runs comfortably at ~18–25 tok/s.
- Workstation: 2× Mac Studio Max 64 GB
- Insufficient on a single system. With llama.cpp RPC swarm and 4× Mac Studio: possible, ~5 tok/s.
#2. Architecture (efficient V4 variant)
V4 Flash is a proportional reduction of V4 Pro: ~5.7× fewer experts, ~3.8× fewer active parameters, but the same CSA+HCA + Muon + FP4/FP8 skeleton. This consistency lets V4 Flash inherit the Pro's quality at a much lower cost.
- Experts per layer
- 128 (vs 256 for Pro). Identical Top-8 routing.
- Active parameters
- 13B (vs. 49B for Pro). Inference speed ~3.8× higher at equivalent total params.
- Layers
- 61 (Pro: 76). Moderate reduction to preserve reasoning depth.
- Attention heads
- Identical to the Pro (CSA+HCA configured identically). No savings on attention to preserve long-context quality.
- Pretraining
- “Efficient reasoning” target: 18T tokens (vs 32T for Pro), with a data mix optimized for math + code.
#3. Hardware by Quantization
| Quantization | Model VRAM | + KV 32k | Recommended configurations |
|---|---|---|---|
| FP16 (reference) | 568 GB | ~620 GB | DC: 4× H200 141GB. Impractical locally. |
| Q8_0 | 305 GB | ~340 GB | Mac Studio M3 Ultra 512 GB or 8× RTX 4090 24GB. |
| Q5_K_M | 205 GB | ~235 GB | Mac Studio Ultra 256 GB (barely enough), 4× RTX A6000 48GB comfortable. |
| Q4_K_M | 170 GB | ~200 GB | Mac Studio Ultra 256 GB, 4× RTX 4090 24GB, 2× RTX 6000 Ada 48GB. |
| IQ3_M (compressed) | 130 GB | ~160 GB | Mac Studio M2 Ultra 192 GB, 4× RTX 4090 (light offload). |
| IQ2_XS (extreme) | 85 GB | ~115 GB | RTX 5090 + 64 GB DDR5 offload, or 2× RTX 4090. |
#4. Mac Studio Ultra 256/512 GB setup
#5. NVIDIA multi-GPU setup
#Option A — vLLM 4× RTX A6000 (48 GB × 4 = 192 GB)
#Option B — llama.cpp with partial offload
#6. Measured benchmarks and tokens/sec
| Benchmark | V4 Flash | V4 Pro | Llama 4 Scout 109B | Mistral Large 2 |
|---|---|---|---|---|
| MMLU-Pro | 79.4 | 84.2 | 76.8 | 75.3 |
| GPQA Diamond | 70.1 | 78.5 | 63.2 | 59.4 |
| AIME 2025 | 76.5 | 87.0 | 61.3 | 52.1 |
| LiveCodeBench v5 | 72.6 | 79.8 | 65.4 | 61.8 |
| LongBench v2 (1M) | 68.9 | 72.6 | — | — |
| Setup | Tok/s mode No | Tokens/s in High mode | Max mode tok/s |
|---|---|---|---|
| Mac Studio M3 Ultra 256 GB | 10-12 | 8-10 | 5-7 |
| Mac Studio M3 Ultra 512 GB | 12-14 | 10-12 | 7-9 |
| 4× RTX A6000 48GB | 20-25 | 16-20 | 12-15 |
| 2× RTX 5090 + 64GB offload | 3-5 | 2-4 | 1-2 |
#7. Thinking modes and long context
V4 Flash inherits V4 Pro's three thinking modes, with slightly lower quality but much lower inference cost. It's the right choice for agents that need to reason without monopolizing infrastructure.
- No Mode
- Direct response. Very low latency (~1s to start streaming). For conversational chat, translation.
- High mode
- 200–2000 chain-of-thought tokens before the answer. +25% quality on math/code. 5–15s latency before the first token of the final answer.
- Max mode
- Up to 16k reasoning tokens. 1-5 minutes of local latency. Reserve it for hard problems (math proofs, complex debugging).
- 1M long context
- Same capabilities as V4 Pro thanks to HCA—Needle-in-Haystack 1M ~95% (vs. 98% for Pro). Excellent for RAG on large databases.
#V4 Flash vs. V4 Pro: choosing
| Criterion | V4 Flash 284B | V4 Pro 1.6T |
|---|---|---|
| Total parameters | 284 B | 1 600 B |
| Active parameters / token | 13 B | 49 B |
| Q4_K_M VRAM | 170 GB | 960 GB |
| Minimum local hardware | Mac Studio Ultra 256 GB (used M3 Ultra or new M5 Ultra) | ~€80,000 cluster |
| Typical local tokens/sec | 10-25 | 0,5-2 |
| MMLU-Pro quality | 79.4 | 84.2 |
| AIME 2025 quality | 76.5 | 87.0 |
| Native context | 1 M | 1 M |
| Max thinking mode | Yes | Yes (better quality) |
| License | MIT | MIT |
#Officialized on July 31, 2026 (build 0731)
DeepSeek released V4-Flash from preview in late July 2026 with build « 0731 »: the API is in public beta, and along the way the launch price of V4-Pro dropped by 75% ( $0.87 per million output tokens). Artificial Analysis places the V4 family "among the open-weight leaders." In terms of weights, nothing changes: 284B total with 13B active — still the only semi-practical V4 variant for high-end local use (256 GB Mac, Strix Halo, multi-GPU) with aggressive quantization.
Since August 2, LM Studio offers it too: locally if your machine can handle it, or through the Bionic app on U.S. servers with “Zero Data Retention” by default—a trade-off worth knowing about, even though this site's philosophy remains 100% local.
#FAQ
Does DeepSeek V4 Flash really run on a Mac Studio M3 Ultra 256 GB?+
How much does a viable local setup for V4 Flash cost?+
Should you choose V4 Flash or wait for a 70B distillation?+
Does V4 Flash beat Llama 4 Maverick / Scout?+
How much power does a Mac Studio M3 Ultra consume during V4 Flash inference?+
Is Max thinking mode worth it on V4 Flash?+
Can you fine-tune V4 Flash locally?+
Is there a stable Q4 quantized version as of April 25, 2026?+
Prices change quickly: every Monday and Thursday, our tracker records the lowest price for local AI graphics cards, along with the price per GB of VRAM.
Feedback, an error, or a clarification? Let us know—it improves the guide for everyone.