Yanis Miraoui retweeted
🚀 Mercury 3 has launched!
🧪 Full benchmarks are now available. Mercury is on par or better than frontier speed-optimized models like GPT-6 Luna, Claude Haiku 5.5, but runs at 1000 tok/sec!
👉 Check out the full table: inceptionlabs.ai/blog/introduci…
🧪 Full benchmarks are now available. Mercury is on par or better than frontier speed-optimized models like GPT-6 Luna, Claude Haiku 5.5, but runs at 1000 tok/sec!
👉 Check out the full table: inceptionlabs.ai/blog/introduci…
5
8
71
4,122
RT @samar_a_khanna: I’ve been waiting to share Mercury 3. >1000 tok/s and a massive step up in quality from our previous diffusion reasonin…
0
3
0
1
Yanis Miraoui retweeted
The most capable diffusion LLM ever released 🧠
Mercury 3 beats GPT-6 Luna and Claude Haiku 5.5 while generating 1,000+ tok/s ⚡️
Try it yourself 👉 chat.inceptionlabs.ai
Mercury 3 beats GPT-6 Luna and Claude Haiku 5.5 while generating 1,000+ tok/s ⚡️
Try it yourself 👉 chat.inceptionlabs.ai
1/4 Today, we’re releasing Mercury 3. Mercury 3 is by far our most powerful model and the most intelligent diffusion LLM ever released. It outperforms GPT-6 Luna and Haiku 5.5 on quality while running at 1000+ tok/sec.
1
4
28
984
Yanis Miraoui retweeted
📢 Mercury 3 is here, the most capable diffusion LLM released to date!
🧠 3x more intelligent than Mercury 1 across standard benchmarks, ahead of GPT-6 Luna and Claude Haiku 5.5
🚀 Over 1,000 tokens/sec, the same speed as Mercury 1 at more than 40x the size
🎯 Early partners are already running it in browser agents and real-time tutoring
Diffusion keeps its speed advantage as it scales, so you no longer have to choose between a fast model and a capable one.
Very proud of the entire Inception team!
🧠 3x more intelligent than Mercury 1 across standard benchmarks, ahead of GPT-6 Luna and Claude Haiku 5.5
🚀 Over 1,000 tokens/sec, the same speed as Mercury 1 at more than 40x the size
🎯 Early partners are already running it in browser agents and real-time tutoring
Diffusion keeps its speed advantage as it scales, so you no longer have to choose between a fast model and a capable one.
Very proud of the entire Inception team!
1/4 Today, we’re releasing Mercury 3. Mercury 3 is by far our most powerful model and the most intelligent diffusion LLM ever released. It outperforms GPT-6 Luna and Haiku 5.5 on quality while running at 1000+ tok/sec.
1
13
127
7,205
Yanis Miraoui retweeted
Mercury 3 is live on OpenRouter!
@_inception_ai's new diffusion model generates 1,000+ tok/s with 1M context, 4x more than Mercury 2.5. Inception reports it beats GPT-6 Luna and Haiku 5.5 on IFBench, GPQA Diamond, and SWE-Bench Atlas.
50% off at launch: $0.10/M in, $0.35/M out
@_inception_ai's new diffusion model generates 1,000+ tok/s with 1M context, 4x more than Mercury 2.5. Inception reports it beats GPT-6 Luna and Haiku 5.5 on IFBench, GPQA Diamond, and SWE-Bench Atlas.
50% off at launch: $0.10/M in, $0.35/M out
48
33
743
46,053
Diffusion scales!
2/4 Mercury 3 demonstrates that text diffusion scales. At over 40x the size of Mercury 1, it’s vastly more intelligent while remaining just as fast. This is just the beginning: diffusion opens a new path to scaling AI, where intelligence can grow without a proportional increase in latency.
1
0
10
350
Yanis Miraoui retweeted
2/4 Mercury 3 demonstrates that text diffusion scales. At over 40x the size of Mercury 1, it’s vastly more intelligent while remaining just as fast. This is just the beginning: diffusion opens a new path to scaling AI, where intelligence can grow without a proportional increase in latency.
5
8
398
37,486
Yanis Miraoui retweeted
3/4 Mercury 3 is already powering diverse agentic applications, including code review, interactive education, and browser use. Early adopters include @Tiny_Fish, Lorikeet, @revisiondojo, Bauplan, @OctenAI, and Croux.
12
16
346
30,239
Better than Haiku and Luna at >1,000 tokens/sec! No more waiting when coding and again a proof that dLLMs can scale
1/4 Today, we’re releasing Mercury 3. Mercury 3 is by far our most powerful model and the most intelligent diffusion LLM ever released. It outperforms GPT-6 Luna and Haiku 5.5 on quality while running at 1000+ tok/sec.
3
2
26
851
Yanis Miraoui retweeted
1/4 Today, we’re releasing Mercury 3. Mercury 3 is by far our most powerful model and the most intelligent diffusion LLM ever released. It outperforms GPT-6 Luna and Haiku 5.5 on quality while running at 1000+ tok/sec.
287
458
5,267
479,258
So so proud of the team!
Excited to share that Mercury 2 is one of @TIME's Best Inventions of 2026.
Not long ago, diffusion LLMs were just a research idea. Now Mercury powers products people use every day.
Proud of the @_inception_ai team. Something bigger coming next week 👇🧵
1
1
15
274
Yanis Miraoui retweeted
Very proud of our team. Mercury 2 has been recognized as one of the best inventions of 2026, alongside massively impactful products like Claude Code. Mercury 2 is also the least pareto-dominant our models will ever be. Better things to come soon... ;)
Excited to share that Mercury 2 is one of @TIME's Best Inventions of 2026.
Not long ago, diffusion LLMs were just a research idea. Now Mercury powers products people use every day.
Proud of the @_inception_ai team. Something bigger coming next week 👇🧵
0
7
24
658
Yanis Miraoui retweeted
Our largest and most capable model lands next week.
If you’re building something where speed matters, we’d love to get it in your hands early:
inceptionlabs.ai/enterprise#con… (2/3)
If you’re building something where speed matters, we’d love to get it in your hands early:
inceptionlabs.ai/enterprise#con… (2/3)
2
1
14
2,724
Yanis Miraoui retweeted
Excited to share that Mercury 2 is one of @TIME's Best Inventions of 2026.
Not long ago, diffusion LLMs were just a research idea. Now Mercury powers products people use every day.
Proud of the @_inception_ai team. Something bigger coming next week 👇🧵
Not long ago, diffusion LLMs were just a research idea. Now Mercury powers products people use every day.
Proud of the @_inception_ai team. Something bigger coming next week 👇🧵
10
25
135
12,105
Yanis Miraoui retweeted
Today was our biggest day ever. Jev had a whole launch day. We quietly grew 155%. #3 and climbing.
1
6
43
3,656
Yanis Miraoui retweeted
Mercury Decide from @_inception_ai now accepts image input on OpenRouter
Put images alongside text in the state of a Decisions request, and Mercury Decide answers your typed questions from what's in the image: a choice, a score, or a yes/no with a probability attached
Pricing is unchanged: free endpoint, or $0.02/M input with zero data retention. Images bill as input tokens
openrouter.ai/inception/merc…
Put images alongside text in the state of a Decisions request, and Mercury Decide answers your typed questions from what's in the image: a choice, a score, or a yes/no with a probability attached
Pricing is unchanged: free endpoint, or $0.02/M input with zero data retention. Images bill as input tokens
openrouter.ai/inception/merc…
6
11
82
9,031
Yanis Miraoui retweeted
1 cent for 669 clinical decisions, and it's tied #2 on my System One board 🔥
Mercury Decide from Inception got 630 right, the cheapest of all 9 models above 90%
you asked for it, so i ran it 😄🏥
Mercury Decide from Inception got 630 right, the cheapest of all 9 models above 90%
you asked for it, so i ran it 😄🏥
Last week, we launched Mercury Decide, a Jev-beating decision model that quickly rose to #2 on OpenRouter by number of users: openrouter.ai/inception/merc…
This week, we’re updating Mercury Decide with substantially better benchmark performance, longer context (33k → 66k), General Data Retention (ZDR), and more.
All that with lower pricing than Jev (ZDR for $0.02/$0 per 1M in/out, discounted at 50% from the list price of $0.04/$0, plus 100M free tokens), alongside an unlimited free tier.
What have we improved? Read on to find out! [🧵1/4]
4
6
51
2,715
Yanis Miraoui retweeted
I tested @typesafeai's jev vs @_inception_ai's decide on my ffl roster.
I simulated 300+ legal lineup combinations along with different context combinations...
Mercury’s projection-informed lineup scored 109.88.
That’s 13.3 more than Jev and 4.3 more than my projection ifier should i try next?
I simulated 300+ legal lineup combinations along with different context combinations...
Mercury’s projection-informed lineup scored 109.88.
That’s 13.3 more than Jev and 4.3 more than my projection ifier should i try next?
0
1
10
776
Yanis Miraoui retweeted
Mercury Decide from @_inception_ai is now available with zero data retention on OpenRouter
The fastest growing decision model on OpenRouter last week now has a paid ZDR endpoint, alongside the free one
$0.02/M input (50% off the $0.04 list price). Output and cached input are s://t.co/5sumvRshSm
The fastest growing decision model on OpenRouter last week now has a paid ZDR endpoint, alongside the free one
$0.02/M input (50% off the $0.04 list price). Output and cached input are s://t.co/5sumvRshSm
11
15
110
11,153