M5 Ultra Mac Studio reviews call it the best machine yet for local AI
First reviews of Apple's priciest Mac Studio, out ahead of Tuesday's launch, focus on its power to run large AI models entirely on-device.
Part of a larger narrative
Apple's AI-Optimized Silicon Push
- M5 Ultra Mac Studio reviews call it the best machine yet for local AIyou are here
- Apple's M6 Mac mini reviews land: faster chip, $300 higher price
What to know
- New Mac Studio (M5 Max from $2,499, M5 Ultra from $5,499, up to $12,299 as tested) ships Tuesday September 22.
- Reviews converge on local AI performance as the standout use case, with reviewers running large models like Qwen3.8-Flash-Next and Deepseek v4 Flash entirely on-device.
- Benchmarks show roughly 30% CPU gains and up to 59-63% GPU gains over the M3 Ultra, with one reviewer noting storage speed as the biggest practical upgrade.
- Price remains a sticking point: even Apple's own reviewers call the top configuration's cost justified only for AI developers and specialized professionals, not typical creators.
The dispute Whether the machine's extreme price is a reasonable trade-off for eliminating cloud AI costs or only makes sense for a narrow set of AI-developer and VFX buyers. · positions read across 37 posts and comments
The M5 Ultra Mac Studio makes local AI agents genuinely fast and practical, rivaling or replacing cloud models.
-
“I reviewed the M5 Ultra Mac Studio with 256 GB of RAM. It’s the dream Mac for local AI agents. And I’ve gone local-only with Hermes.”
viticci@macstories.net · Mastodon ↗
Its raw benchmark power is impressive even outside AI, shifting who the machine is really built for.
-
“Apple’s $12,300 Mac Studio with an M5 Ultra chip and 256GB of RAM is outrageously powerful”
Antonio G. Di Benedetto · Mastodon ↗
“At $11,299, the M5 Ultra's extra performance over the M5 Max model is only worth it for select buyers.”
Engadget, Reviewer · Engadget ↗ · Sep 20
Federico Viticci MacStories editorApple ManufacturerAntonio G. Di Benedetto The Verge reviewerMarques Brownlee (MKBHD) Tech reviewer/YouTuberRoman Loyola Macworld reviewer
How it unfolded 6 developments, newest first · click a bar or a number to jump articlespostscomments
-
6
MKBHD says the machine now targets AI users over creators
Reviewer Marques Brownlee posts that after a week of testing, the M5 Ultra Mac Studio's improvements now seem aimed at local AI workloads rather than video creators like himself.
“At this point, the improvements are not even targeted at me (a video creator) anymore. They're targeted at being the best machines for local AI work.”
— @MKBHD -
Been using the M5 Ultra Mac Studio for the past week and predictably it is the most powerful computer I've ever used. At this point, the improvements are not even targeted at me (a video creator) anymore. They're targeted at being the best machines for local AI work. There's
2 more of the top 3 · 26 posts in this stretch
-
The Cost of a 27- inch fully loaded iMac from Apple $3,700 in 2011 is worth $5,510.05 today (fully loaded) used for 10 years.The Mac Pro Tower plus the Cinema Display monitor at the time was even more, similar to the Studio M5 ULtra today than the iMac which I believe was an upper middle computer?$5,700 in 2011 is worth $8,488.46 today$7,700 in…
-
P
Apple's 2026 Mac Studio with the M5 Ultra pushes performance (and pricing) to the limit. The new quad-die chip is a serious powerhouse for top-tier local AI inference, heavy software compilation, and high-throughput video rendering. Here's our review: https://www. pcmag.com/reviews/apple-mac-st udio-2026-m5-ultra
-
-
5
MacRumors flags storage speed as the real upgrade
MacRumors highlights Macworld's review finding that the new Mac Studio's storage speed gains, not the M5 Ultra chip itself, stood out most in testing.
“The personal assistants I use the most are now entirely powered by a model running locally on a Mac Studio.”
— Federico Viticci, MacStories editor · source -
first by 9to5Mac, 2d ago · also MacRumors, MacRumors Forums, HN Best, HN Frontpage, MacStories
3 more headlines
- New Mac Studio's Biggest Upgrade Isn't the M5 Ultra Chip, Review Says MacRumors · 2d ago
- M5 Ultra Mac Studio Review HN Best · 2d ago
- M5 Ultra Mac Studio Review: The Dream Mac for Local AI Agents HN Frontpage · 2d ago
-
I ran all the M5 Ultra tests I could think of: - Two models (Flash-Next and GLM-5.3-Flash) - 4k, 8k, 16k, 64k, 128k, and 256k prompts - M5 Ultra vs M3 Ultra vs RTX 5090 - 4/5/6/8-bit Flash-Next quants I created all the possible charts and visualizations I could think of. Hope they’re useful! https://www. macstories.net/stories/m5-ultr…
-
-
4
MacStories names M5 Ultra Studio the dream local-AI Mac
Federico Viticci's MacStories review reports switching his personal AI assistants entirely to local models (Qwen3.8-Flash-Next) on the M5 Ultra Studio, citing faster GPU throughput, higher memory bandwidth and much faster token prefill than the M3 Ultra.
-
This M5 Ultra Mac Studio feels made for Qwen3.8-Flash-Next. - 100+ TPS at 4 or 5-bit - Token prefill is up 150% (!) compared to M3 Ultra - Can use subagents with room to spare Flash-Next feels GREAT for agentic coding, especially at these speeds. https://www. macstories.net/stories/m5-ultr a-mac-studio-review-the-dream-mac-for-local-ai-agents/
2 more of the top 3 · 8 posts in this stretch
-
Been using the M5 Ultra Mac Studio for the past week and predictably it is the most powerful computer I've ever used. At this point, the improvements are not even targeted at me (a video creator) anymore. They're targeted at being the best machines for local AI work. There's nothing I can make Fin...
-
I reviewed the M5 Ultra Mac Studio with 256 GB of RAM. — It's the dream Mac for local AI agents. And I've gone local-only with Hermes. — My story, with LOTS of interactive charts: — www.macstories.net/stories/m5- u...
-
-
3
Embargo lifts on Mac Studio M5 Max/Ultra reviews
A wave of first reviews from The Verge, Engadget, Macworld, CNET, Tom's Hardware, Tom's Guide, PCMag and MacRumors publishes simultaneously ahead of the new Mac Studio's Tuesday launch, covering the M5 Max ($2,499+) and M5 Ultra ($5,499+, up to $12,299 as tested) configurations.
“Apple’s $12,300 Mac Studio with an M5 Ultra chip and 256GB of RAM is outrageously powerful…”
— Antonio G. Di Benedetto -
first by WinCentral, 2d ago · also Geek Culture, PCMag, Tom's Hardware, Macworld
4 more headlines
- Apple Mac Studio (M5 Max, 2026) - Review Geek Culture · 2d ago
- Apple Mac Studio (2026, M5 Ultra) Review: Quad-Die Processing Power to Die For PCMag · 2d ago
- Apple Mac Studio (M5 Ultra) review: Local model citizen outpaces DGX Spark and Threadripper Tom's Hardware · 2d ago
- Mac Studio (M5 Max) review: Shifting into overdrive Macworld · 2d ago
-
first by iThinkDifferent, 2d ago · also BGR, AppleInsider, Engadget
4 more headlines
- Mac Studio (M5 Ultra) Review: The Ultimate AI Workstation BGR · 2d ago
- M5 Ultra Mac Studio's GPU, AI performance keeps up with desktop graphics cards AppleInsider · 2d ago
- Mac Studio review roundup: expensive, but M5 Ultra makes it worthwhile AppleInsider · 2d ago
- Apple Mac Studio (M5 Ultra) review: incredible benchmark scores, high base memory, and a generous number of ultra-fast Thunderbolt 5 ports, but very expensive Engadget · 2d ago
-
I reviewed the M5 Ultra Mac Studio with 256 GB of RAM. It’s the dream Mac for local AI agents. And I’ve gone local-only with Hermes. My story, with LOTS of interactive charts: https://www. macstories.net/stories/m5-ultr a-mac-studio-review-the-dream-mac-for-local-ai-agents/
1 more of the top 2 · 2 posts in this stretch
-
I Tested the New M5 Apple Mac Studio, a Pricey, Quad-Die Beast for Heavy AI Use https://www. pcmag.com/reviews/apple-mac-st udio-2026-m5-ultra?utm_source=flipboard&utm_medium=activitypub Posted into Desktop PCs: Expert Reviews and Top Deals @ desktop-pcs-expert-reviews-and-top-deals-PCMag
-
- 2 days quiet
-
2
Leaked GPU benchmarks show M5 Ultra graphics gains
GPU benchmark results surface for the M5 Ultra and M6 chips, with the M5 Ultra scoring up to 59% higher on Geekbench 7 Metal than the M3 Ultra and landing near Nvidia's RTX 4080 on OpenCL.
-
first by MacRumors, 5d ago
-
first by TechPowerUp, 5d ago
-
- 2 days quiet
-
1
First Geekbench leaks reveal M5 Ultra CPU gains
Early Geekbench CPU benchmark results for the M5 Ultra surface online, with coverage noting large multi-core gains over rival high-core-count CPUs like AMD's Threadripper PRO.
What people are saying 17 voices from 3 sites · best of 37 · verbatim
- Does running local models actually save money over time compared to paying for top cloud AI subscriptions?
- Sep 22
-
N
M5 Ultra Mac Studio Review: https://www. macstories.net/stories/m5-ultr a-mac-studio-review-the-dream-mac-for-local-ai-agents/ Discussion: http:// news.ycombinator.com/item?id=4 9787313
- Sep 21
-
J
https://www. macstories.net/stories/m5-ultr a-mac-studio-review-the-dream-mac-for-local-ai-agents/ This was pretty interesting!
-
A basic llama-bench on Qwen 3.8 27B UD-Q4_K_M gives pp512 3920 tok/s / tg128 81 tok/s on a 500W RTX PRO 6000 (should be similar speeds to a 5090, chip is basically the same, just less VRAM). With MTP3 this is 140 tok/s on mtp-bench.This is with llama.cpp. You can of course use vLLM/SGLang well on these cards and they're even faster. On vLLM w/…
-
I really appreciate seeing these dense model numbers. For a large unified memory system though I expect that MoE numbers are what people are more interested in.These numbers could and should get much better. As an example I can run Qwen3.8-27B-MXFP4 (W4A8) on 2x AMD R9700 that gets 260+ tokens/sec to start and slows down to ~110 tokens/sec over…
-
Once you hit the memory you need, generation speed is mainly set by bandwidth, and every Ultra from M1 thru M3 has ~800 GB/s. IMO best ROI for most people is 'cheapest used Ultra with enough RAM'I setup an eBay alert and picked up a used M2 Ultra that has delivered good ROI (at least, far better than 15k for comparable-for-my-use-case performance)
-
On my m5 max 27b model does 75tps on 256k ctx and starts at 80 on the 8k ctx when you add https://huggingface.co/collections/z-lab/dflash-2 to it. So yeah base might be 30tps (I used iq4) but mtp or dflash help a lot and should be used when checking what is useful and what is not for running models as it is not fare to judge without them.
-
My mac is 5 years old. I don't think I can comfortably buy a new one right now. It has a 16GB unified RAM. Honestly that would be enough for so many local models that I want to use but can't use. Because RAM usage (even with literally every single user installed app quit/stopped) the RAM usage is very high that I can barely safely get 6-7 GB (I am…
-
Those are some incredible graphs, that leap in prompt processing going from M3 to M5.Also: ~30 token/s on GLM 5.3-flash, locally. (That's roughly Opus 4.8-tier. I think)./meta Here's a CSS filter that stops those nuisance chart animations, macstories.net##*:style(animation: none !important; transition: none !important)
-
While I know it's not apples to apples, the target comparison right now is 2x DGX Sparks. Similar price, 256gb. The conversation has focused on memory bandwidth vs. compute in agentic loops, so for most people the raw numbers will mean less than the "time per task" in coding benchmarks.This is a great article and bodes well for the M5, but we…
-
This is great as a first look, but the author is not a developer, so we don't yet know whether a dev can be as productive with local models on M5 Mac Studio compared to a 20x subscription plan.I'm also curious about any new low hanging optimization opportunities in the kernels for this new hardware.It's already clear to me that M5 Mac Studio is…
-
> "It also happens to be a Mac, with an operating system that looks nice and doesn’t suck"Yes Apple has some of the best hardware out there, albeit overpriced. But the software is such a hindrance and I can't take anyone that states otherwise seriously. If only it had proper Linux support (and the Asahi people do an amazing job but you can…
-
The numbers I was most interested in are tucked away in a chart towards the bottom - the speed comparison of the Mac Studios v.s. a RTX 5090: Qwen3.8 27B tokens/sec generation speed Prompt size 8K 64K 128K 256K RTX 5090 PC 59 51 44 n/a M5 Ultra 48 39 32 24 M3 Ultra 31 23.5 20 15 A whole bunch more comparison numbers in this section:
-
Federico Viticci has an exhaustive review of the new Mac Studio with M5 and 256 GB of RAM. Lots of cool charts on performance. It sounds like a great Mac. The problem is that config will set you back $9k! For me, a $100/month subscription for frontier AI over many years is still better.
-
We got an early unboxing of the 2026 Mac Studio, and Apple packed serious internal upgrades into this compact workstation. 🤖🚨 Powered by either the M5 Max or the new M5 Ultra chip with up to a 36-core CPU and 80-core GPU, it delivers up to 4.3x faster local AI performance and significantly faste...
-
I have been using the Mac Studio with M5 Ultra for the past week, and it's the most powerful computer I've used. Been a bit busy over the past week so my report will be slightly delayed, but just know... Qwen 3.8 Flash Next is the best model for the 256GB version, it rips
-
The M5 Ultra Mac Studio is INSANE! Apple sent me the maxed out version, with a 36-core CPU, 80-core GPU and 256GB of RAM. I used it to run Deepseek v4 Flash (156GB model) fully on device to build a mini-game, which it did in just a few minutes! #MacStudio #M5Ultra @apple
-
The new M5 Ultra Mac Studio is a local AI beast. @viticci tested a 256GB model and has the charts and practical everyday use cases to back it up on MacStories: