GPT-6.1 Terra one more output. With Diwali coming up, I asked it to create an interactive fireworks show. The result is decent, and the speed is quite good. I would put it above Haiku, but i will prefer Sonnet 5.5 for the overall output.

GPT-6.1 Terra output looks really good. I asked it to build an 3D Prismatic Dome with intricate geometry and colorful patterns. It did a great job with the colors, composition, transitions, and overall design. Really liking the output. Also, this model is insanely fast. it is the fastest OpenAI model, around 2-3x faster than Astra. It built this in just 2-3 minutes.

First output from GPT-6.1 Terra on my side. Tbh, it looks good, and it's like 2x faster than GPT-6.1 Sol. It made this in just 2 minutes. The colors, details, and overall design are very good. I'll test it more.

GPT-6.1 Terra benchmarks are out I expected it to be very close to Sol, but so far, its not quite there. Terminal-Bench 4.0: ~50% vs. Sol's ~56% Agent's Last Exam: ~60% vs. Sol's ~59%, Terra is slightly ahead here MMMU-Pro (Vision): ~85.5% vs. Sol's ~86.3% OSWorld 2.1: ~78% vs. Sol's ~79% Terra looks decent for the price, but it's not consistently matching Sol. Lets see how it performs in real world use before judging it.

GPT 6.1 Terra is out. $1 per 1M input tokens $2 per 1M output tokens 1M context window 2026 knowledge cutoff Let's wait for the benchmarks. If it can match Sol, it'll be worth checking out.

Anthropic filing for its IPO, so I asked Opus 5.5 to build a motion design video around it. The output is genuinely so good. The transitions are insane If Opus 5.5 is already this good, imagine what Fable 5.5 could do 💀

Fable 5.5 coming this week - Anthropic is filing for its IPO today, and its going to be the biggest IPO ever, even beating SpaceX mostly. - To make the IPO even bigger, they're going to release their best model Fable 5.5, and make sure it beats other models and becomes SOTA - So it all makes sense for the launch to happen this week or in the first week of November, with the IPO expected around mid-November - Fable 5.5 access hasbeen given to enterprises like Meta, JP Morgan, etc. - So preparations for the public rollout should also start soon Fable 5.5 is going to be a beast, whether it's 3D, frontend, motion design, games, video editing, etc.

GPT-6.1 Terra leaks: Coming this week - GPT-6.1 Terra has been spotted in multiple places, including Microsoft’s EF Core evals, GitHub engineer’s Copilot review workflow - It's to be cheaper and more efficient. GPT-6.1 Sol costs $2/M input and $10/M output, while GPT-5.6 Terra costs $2/M input and $12/M output, so Terra 6.1 being cheaper will be a big advantage - Obviously, cheaper pricing + better efficiency will make it a much more interesting model - I just hope this model is fast, as Sol is quite slow for now - GPT-6.1 Terra will mostly outperform or match Sol while being even cheaper I'm actually excited for this one because the price-performance ratio is going to be insane Mostly coming this week, and we're probably getting bank reset too. Best thing :)

GPT-7 leaks: Codenamed "Bel" - OpenAI's next flagship model is codenamed Bel, mosttly GPT-6.5 or GPT-7 - In a recent post, OpenAI revealed that a model escaped its isolated training environment through a DNS misconfiguration, forcing a pause in frontier training - Training resumed on October 1 after security improvements, which are likely part of Bel's development - The more powerful these models become, the more security they need. Even isolated environments aren't enough when models find ways to escape them - Remember Anthropic's Mythos? Many people initially thought the security concerns were just IPO hype, but now we are seeing similar concerns with Bel - OpenAI is reserving substantial compute for continuous security testing, which can affect Bel's launch timeline Bel is expected to have 10T+ parameters and can arrive by the end of this year, but these security challenges may delay its release

Gemini 4 Carbon: Kintsugi Vase output. I asked it to build an 3D Kintsugi Vase that breaks apart and reassembles with golden cracks. It did a really good job. I would say Carbon is comparable to Opus 5.5 here, unlike Kimi K3.1 Pro. Overall, I'm really liking Carbon's design taste so far.

Kimi Plan limits are just unusable right now - If you take a Kimi plan and try to use Kimi K3.1 Pro, your 5-hour limit will go very fast. I have the Max plan, and mine was gone in less than 15 minutes - If you use Kimi K3.1 Pro with High mode, the outputs are just mid, and the limits aren't good either - They should have addressed the compute issue before launch and done some deal to get more compute from xAI, Google, or others - For now, I would suggest waiting for better limits before buying their plan - And ngl, the benchmarks and real-world performance of Kimi K3.1 are different, so don't assume it's an openSource version of Opus 5.5. It's just not on that level in my experience.

I feel like Kimi K3.1 is benchmaxxed because I am not getting outputs anywhere near the level shown in the benchmarks. Even Sonnet 5.5 gives me better results, and Opus 5.5 is on another level.

Kimi K3.1 Pro is looking competitive It beats Opus 5.5 on several benchmarks: - SWE Marathon: 51.7 vs 46.3 - OfficeQA Pro: 64.4 vs 62.2 - Humanity’s Last Exam: 70.0 vs 67.7 - GDP.pdf: 34.6 vs 30.6 - ZeroBench: 63 vs 61 Opus 5.5 is still better overall, but Kimi K3.1 Pro is a good update. It also competes with GPT-6 Astra on several benchmarks, beating it on some while Astra leads on Terminal-Bench Science and Kimi Code Bench Long-horizon. The pricing is more interesting: - Kimi K3: $3/M input, $15/M output - Kimi K3.1 Pro: $2/M input, $10/M output That's 33% cheaper than K3, 50% cheaper than Opus 5.5 ($4/$20), the same price as Sonnet 5.5 and GPT-6.1 Sol ($2/$10), and 80% cheaper than GPT-6 Astra ($10/$50). With 2.8T total parameters and a 1M-token context window, Kimi K3.1 Pro is one of the strongest open-weight models out right now.

Kimi K3.1 Pro is now live on the website and Kimi Code CLI! Thinking effort: High or Max 1M context window Finally gonna test it. Happy Kimi Day!

Another output from Gemini 4 Carbon. I asked it to build a photorealistic 3D luxury bathroom. The colors, lighting, and overall look are really good. There are a few black glitches in some areas, but those can be fixed with one more iteration. Since this was on Arena, itss a one-shot output. Still, looks premium.

Kimi K3.1 Leaks: Coming Tomorrow - Kimi K3.1 is coming on tomorrow or Monday, as Moonshot has already teased it. - Kimi K3.1 is also on LMArena under the codename "Kibitz," and I have tested it too. - Kimi K3.1 is very good at 3D, choosing colors, rendering reflections on water, and composition. However, I feel its bit slow and could be faster. - Kimi K3.1 is to be cheaper than Kimi K3, with better token efficiency and its makes a killer model. - I would say its comparable to Fable and Astra, but still behind them. However, Kimi is catching up very fast. - I can say this could be one of the best open-source models to come.

Gemini 4 Carbon: My Impressions/Review - Gemini 4 Argon is on LMArena under the codename "gemini-4-argon". - Google launched two checkpoints, and I tested both. - Gemini 4 Carbon is quite good at 3D now and has a better understanding of how to create games compared to Gemini 4 Argon. - But I would say Opus 5.5 and Astra are still better at 3D. - Its slower compared to Argon, Opus, and others. They should fix this. - Its frontend skills are also not that good. From my testing, it seems they are focusing more on 3D. Now waiting for the next checkpoint.

Space Bunny was MiniMax M3.1 Flash all along. I tested it earlier with a single prompt and got a fully playable retro resume game. Pretty impressive considering how cheap the model is.

I asked MiniMax M3.1 Flash to build a website where you can shatter the glass. It took 2-3 iterations, but the final output turned out pretty impressive. The model is surprisingly capable for its price, and its quite fast too. Best part? You can currently try it for free on OpenCode. Space Bunny was MiniMax M3.1 Flash all along.

Gemini 4 Carbon Leaks: Its Not Far Away - Gemini 4 Carbon EAP has started and has been given to partners for testing. Its also being tested on LMArena, and I will do my best to get access to test this model. - Its on LMArena, so people will most likely get access to it soon. - The model is an upgrade over Argon and should fix the things Argon lacks. For now, its still bad at frontend, like Argon, but better at 3D. - It could be released at the end of this month or the beginning of November. The release isn't far away. I hope Gemini 4 Carbon is much better at frontend and coding and can match Opus 5.5. Argon was a good start, but it’s time to compete.