Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Kimi K3 ranks #6 on ReactBench - It performs better than Opus 4.8 while being over 2x cheaper (!!) - Scores 33%, jump of +10% compared to K2.7 - Lags behind the frontier (Fable and Sol) in frontend production-readiness:

aidenybai's profile picture

Aiden Bai

70,402 subscribers

75,947 görüntüleme • 3 ay önce •via X (Twitter)

Bilim & Teknoloji
LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

I tested Kimi K3 vs Claude Opus 4.8 Same prompt, an armory bay with lighting, props, and detail. Top is Kimi K3, bottom is Opus 4.8. It's not even close. Kimi K3 built a full scene with textures, proper lighting, ammo crates, weapon racks, working detail everywhere. Opus 4.8 gave me a near empty room with a couple of floating tables. No doubt it beats Opus 4.8. Kimi K3 is Fable 5 level, and it's clearly better than GPT-5.6 Sol at 3D and games. An open weight model just matched the best closed models on the market. Let that sink in.
0:13

I tested Kimi K3 vs Claude Opus 4.8 Same prompt, an armory bay with lighting, props, and detail. Top is Kimi K3, bottom is Opus 4.8. It's not even close. Kimi K3 built a full scene with textures, proper lighting, ammo crates, weapon racks, working detail everywhere. Opus 4.8 gave me a near empty room with a couple of floating tables. No doubt it beats Opus 4.8. Kimi K3 is Fable 5 level, and it's clearly better than GPT-5.6 Sol at 3D and games. An open weight model just matched the best closed models on the market. Let that sink in.

Bhavy☄️

481,139 görüntüleme • 3 ay önce

Kimi-K3 VS Opus-4.8 GTA IV test. Left : Kimi, Right : Opus-4.8 Kimi-K3 got impressively better than its previous model. + Muse Spark 1.1 generated something that can’t be played at all. + Gemini 3.5 Flash called itself from Anthropic and the game can’t be played too.
0:14

Kimi-K3 VS Opus-4.8 GTA IV test. Left : Kimi, Right : Opus-4.8 Kimi-K3 got impressively better than its previous model. + Muse Spark 1.1 generated something that can’t be played at all. + Gemini 3.5 Flash called itself from Anthropic and the game can’t be played too.

Jun Song

26,335 görüntüleme • 3 ay önce

🤯KIMI K3 ABSOLUTELY MOGS! BEATING Opus 4.8, GPT 5.5, and even Fable 5 in multiple benchmarks. They scored 1688 on GDPval-AA v2 🔥 This is a completely different breed of open-source models Kimi creates better games and front-end designs than Fable 5, but it's 8x cheaper! The results coming out are truly impressive! I will be testing this out further and posting multiple tests today. Stay tuned!
0:24

🤯KIMI K3 ABSOLUTELY MOGS! BEATING Opus 4.8, GPT 5.5, and even Fable 5 in multiple benchmarks. They scored 1688 on GDPval-AA v2 🔥 This is a completely different breed of open-source models Kimi creates better games and front-end designs than Fable 5, but it's 8x cheaper! The results coming out are truly impressive! I will be testing this out further and posting multiple tests today. Stay tuned!

Mark Santos

128,273 görüntüleme • 3 ay önce

GPT-6 Astra vs Fable 5.1 vs Kimi K3 vs SOL Astra is a clear jump from SOL. It understands intent really well. These models are getting so powerful, it’s hard to pick a winner. Fable is great at realism, while Kimi K3 has improved a lot since launch. Which one do you like?
0:17

GPT-6 Astra vs Fable 5.1 vs Kimi K3 vs SOL Astra is a clear jump from SOL. It understands intent really well. These models are getting so powerful, it’s hard to pick a winner. Fable is great at realism, while Kimi K3 has improved a lot since launch. Which one do you like?

Bhavy☄️

42,313 görüntüleme • 2 ay önce

Kimi K3 vs Claude 4.8 Opus colosseum! 2.50$ vs 7.60$ Which one did better?
0:12

Kimi K3 vs Claude 4.8 Opus colosseum! 2.50$ vs 7.60$ Which one did better?

Chris

55,066 görüntüleme • 3 ay önce

Kimi K2.7 Code just made Kimi K2.6 painfully outdated we tested Kimi K2.7 Code against Kimi K2.6, GPT 5.5, and Claude Opus 4.8 across Lorenz attractors, solar systems, and water waves in our previous comparison, Kimi K2.6 struggles with physical motion, now K2.7 gave the most realistic rendering of water wave!
0:15

Kimi K2.7 Code just made Kimi K2.6 painfully outdated we tested Kimi K2.7 Code against Kimi K2.6, GPT 5.5, and Claude Opus 4.8 across Lorenz attractors, solar systems, and water waves in our previous comparison, Kimi K2.6 struggles with physical motion, now K2.7 gave the most realistic rendering of water wave!

GMI Cloud

87,380 görüntüleme • 4 ay önce

A quick cost per task comparison of Kimi K3 Kimi.ai and GPT-5.6 Sol by OpenAI. At 3.2x cheaper, Kimi K3's model outputs web apps not just on par for design quality, but better across certain domains. Kimi K3 establishes a new design SOTA and Pareto frontier for cost for design tasks.
0:12

A quick cost per task comparison of Kimi K3 Kimi.ai and GPT-5.6 Sol by OpenAI. At 3.2x cheaper, Kimi K3's model outputs web apps not just on par for design quality, but better across certain domains. Kimi K3 establishes a new design SOTA and Pareto frontier for cost for design tasks.

Design Arena

22,584 görüntüleme • 3 ay önce

Snowboarder one-shot on Kimi K3 (OpenCode, Godot). Almost no visual bugs, task cleanly implemented. Better than Sol 5.6 and Opus here. Kimi.ai
0:19

Snowboarder one-shot on Kimi K3 (OpenCode, Godot). Almost no visual bugs, task cleanly implemented. Better than Sol 5.6 and Opus here. Kimi.ai

Alex Ermolov

66,769 görüntüleme • 3 ay önce

DeepSeek V4 Flash being ranked almost equal to Opus 4.8 is actually insane to me. Against Kimi K3, Opus 5, GPT-5.6 Sol, and Qwen 3.8 Max Preview, you give up a LOT by choosing DeepSeek. This is not a 2-point difference in practice. I’ll post the Opus 4.8 comparison next because you need to see this.
0:19

DeepSeek V4 Flash being ranked almost equal to Opus 4.8 is actually insane to me. Against Kimi K3, Opus 5, GPT-5.6 Sol, and Qwen 3.8 Max Preview, you give up a LOT by choosing DeepSeek. This is not a 2-point difference in practice. I’ll post the Opus 4.8 comparison next because you need to see this.

OmedTheVibeCoder

26,990 görüntüleme • 3 ay önce

I gave Kimi K3 and Opus 5 the exact same prompt to build a car racing game. I expected Opus 5 to take the lead. Instead, Kimi K3 finished in ~25 min for ~$8.9, while Opus 5 took 19 min and cost $14. The biggest surprise wasn’t the speed or the price. Kimi K3 built the better game. Smoother controls, better physics, and a driving experience that felt ready from the start. Opus 5 looked cleaner visually but Kimi won where it actually mattered. Nearly half the cost and the better gameplay.
0:16

I gave Kimi K3 and Opus 5 the exact same prompt to build a car racing game. I expected Opus 5 to take the lead. Instead, Kimi K3 finished in ~25 min for ~$8.9, while Opus 5 took 19 min and cost $14. The biggest surprise wasn’t the speed or the price. Kimi K3 built the better game. Smoother controls, better physics, and a driving experience that felt ready from the start. Opus 5 looked cleaner visually but Kimi won where it actually mattered. Nearly half the cost and the better gameplay.

Aman

112,345 görüntüleme • 3 ay önce

> be Kimi Founder > Zhilin Yang. nobody in the West knows your name. > build K1. K2. K2.5. K2.6. K2.7. > each one sets the open-source record > everyone says China can't reach the frontier > raise $2B. hit $20B valuation. > keep your head down. keep shipping. > July 16. drop K3. > 2.8 trillion parameters. > the biggest open model ever released. > 1 million token context. native multimodal. > Kimi Delta Attention — 6.3x faster decoding. > Artificial Analysis ranks it #4 overall. > right behind GPT-5.6 Sol and Fable 5. > open weights July 27. free for anyone. > the closed labs charge for the frontier. > you gave it away. > different game.
0:14

> be Kimi Founder > Zhilin Yang. nobody in the West knows your name. > build K1. K2. K2.5. K2.6. K2.7. > each one sets the open-source record > everyone says China can't reach the frontier > raise $2B. hit $20B valuation. > keep your head down. keep shipping. > July 16. drop K3. > 2.8 trillion parameters. > the biggest open model ever released. > 1 million token context. native multimodal. > Kimi Delta Attention — 6.3x faster decoding. > Artificial Analysis ranks it #4 overall. > right behind GPT-5.6 Sol and Fable 5. > open weights July 27. free for anyone. > the closed labs charge for the frontier. > you gave it away. > different game.

Kirill

413,564 görüntüleme • 3 ay önce

Kimi K3 just jumped 17 places to #1 on Arena.ai's Frontend Code Arena. So we gave it real client work to see where leaderboards meet reality: >10 landing page briefs, >judged blind by 8 working designers >against GPT 5.6 Sol, Gemini 3.5 Flash, and Claude Fable 5. It tied the best model in the study. And on detailed briefs, it won. Details in 🧵:
0:15

Kimi K3 just jumped 17 places to #1 on Arena.ai's Frontend Code Arena. So we gave it real client work to see where leaderboards meet reality: >10 landing page briefs, >judged blind by 8 working designers >against GPT 5.6 Sol, Gemini 3.5 Flash, and Claude Fable 5. It tied the best model in the study. And on detailed briefs, it won. Details in 🧵:

ben

13,798 görüntüleme • 3 ay önce

🚨 Kimi K3.1 is expected tomorrow, possibly Monday. The best part is it’s CHEAPER than Kimi K3 while being more token-efficient. I’ve been using it frequently over the past week, and it’s been good at Three.js. This demo tests its reverse-engineering capabilities.
0:14

🚨 Kimi K3.1 is expected tomorrow, possibly Monday. The best part is it’s CHEAPER than Kimi K3 while being more token-efficient. I’ve been using it frequently over the past week, and it’s been good at Three.js. This demo tests its reverse-engineering capabilities.

Chris

32,103 görüntüleme • 12 gün önce

Just open sourced MotionPromptGallery 178 AI motion graphics, each with the exact prompt behind it. made with Opus 5.5, GPT-6 Astra, Kimi K3, Fable 5. built with Next.js, fully static, deploys to Vercel in seconds. go ship with it, and add your own piece. live site:
0:15

Just open sourced MotionPromptGallery 178 AI motion graphics, each with the exact prompt behind it. made with Opus 5.5, GPT-6 Astra, Kimi K3, Fable 5. built with Next.js, fully static, deploys to Vercel in seconds. go ship with it, and add your own piece. live site:

MIKE

23,708 görüntüleme • 26 gün önce

KIMI K3 VS OPUS 4.8 SIMULATING A 3D PARTICLE SYSTEM same prompt, same target: thousands of particles reacting to gravity and to each other in real time → K3: more organic distribution, smoother motion, particles actually cluster and drift like something physical → Opus: more rigid pattern, less natural, movement reads more like a grid than a simulation K3: $0.65 in API. Opus 4.8: ~$1.30 this is simulated physics, not just aesthetics, and it's the kind of detail that separates a demo from something you'd actually ship
0:14

KIMI K3 VS OPUS 4.8 SIMULATING A 3D PARTICLE SYSTEM same prompt, same target: thousands of particles reacting to gravity and to each other in real time → K3: more organic distribution, smoother motion, particles actually cluster and drift like something physical → Opus: more rigid pattern, less natural, movement reads more like a grid than a simulation K3: $0.65 in API. Opus 4.8: ~$1.30 this is simulated physics, not just aesthetics, and it's the kind of detail that separates a demo from something you'd actually ship

Jade

32,766 görüntüleme • 3 ay önce

Kimi K3 VS Claude Fable 5. Prompt in the 🧵, now try it on Flowith.
0:14

Kimi K3 VS Claude Fable 5. Prompt in the 🧵, now try it on Flowith.

Flowith

73,302 görüntüleme • 3 ay önce

To show you how beast of a model is opus 5 this sol 5.6 against it that wall where opus is shoooting great wall of china < kimi k3 , qwen 3.8 > we are so back , wait for mac os code to compare with kimi k3
0:11

To show you how beast of a model is opus 5 this sol 5.6 against it that wall where opus is shoooting great wall of china < kimi k3 , qwen 3.8 > we are so back , wait for mac os code to compare with kimi k3

Chetaslua

171,876 görüntüleme • 3 ay önce

Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family. It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.
0:20

Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family. It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.

Claude

29,607,722 görüntüleme • 1 ay önce

Working on a new skill where Fable acts as an orchestrator and delegates work to OpenCode subagents running OSS models. It delegates to GLM 5.3 Flash for easier tasks and Kimi K3 for harder tasks, all running on Together AI. GLM 5.3 Flash is 100x cheaper than Fable for output tokens while being pretty good at coding so I'm excited to experiment with how well this works.
0:14

Working on a new skill where Fable acts as an orchestrator and delegates work to OpenCode subagents running OSS models. It delegates to GLM 5.3 Flash for easier tasks and Kimi K3 for harder tasks, all running on Together AI. GLM 5.3 Flash is 100x cheaper than Fable for output tokens while being pretty good at coding so I'm excited to experiment with how well this works.

Hassan

35,466 görüntüleme • 2 ay önce

GPT-6 Astra vs Fable 5.1 vs Kimi K3 vs Sol Astra: Took 11 minutes. This is from a follow-up after the first attempt had rendering and UI bugs. The smoke feels a little off. I expected more detail, but it did a good job understanding the intent. Cost: $12.85. Fable 5.1: Took 18 minutes to complete the same test with high detailing. Cost: $12.60. Kimi K3: Took 10 minutes with a similar level of output and cost only $6.95. SOL: Took 7 minutes, but the output seemed messy. Cost: $5.78. Personally, I like Kimi K3 for delivering similar quality with better token efficiency, followed by fable 5.1 for realism. What about you? I’ll do a couple more tests before coming to a conclusion.
0:19

GPT-6 Astra vs Fable 5.1 vs Kimi K3 vs Sol Astra: Took 11 minutes. This is from a follow-up after the first attempt had rendering and UI bugs. The smoke feels a little off. I expected more detail, but it did a good job understanding the intent. Cost: $12.85. Fable 5.1: Took 18 minutes to complete the same test with high detailing. Cost: $12.60. Kimi K3: Took 10 minutes with a similar level of output and cost only $6.95. SOL: Took 7 minutes, but the output seemed messy. Cost: $5.78. Personally, I like Kimi K3 for delivering similar quality with better token efficiency, followed by fable 5.1 for realism. What about you? I’ll do a couple more tests before coming to a conclusion.

Bhavy☄️

78,441 görüntüleme • 2 ay önce