Name cannot be blank

10% 🎉🎉 #Bitcoin

@ml_angelopoulos It says Gemma 3 in the picture

Today, Google released the #1 open-source model from a US lab. GEMMA

@testingcatalog @tbpn What no way

BREAKING 🚨: OpenAI aquired @tbpn daily live show! “It’s one of the places where the conversation about AI and builders is actually happening day to day. A lot of you already watch it, and rely on it to stay close to what’s going on.”

@ArtificialAnlys @GoogleDeepMind In the image it says Gemma 4 32b not 31

Google has released Gemma 4, a new family of multimodal open-weight models including Gemma 4 E2B, Gemma 4 E4B, Gemma 4 31B and Gemma 4 26B A4B @GoogleDeepMind’s new Gemma 4 family introduces four multimodal models supporting text, image, and video inputs. We evaluated Gemma 4 31B (dense) and Gemma 4 26B A4B (MoE), both with a 256k context window, while the other two smaller models support up to 128k. With 31B and 26B parameters respectively, both evaluated models can run on a single H100. On GPQA Diamond, our scientific reasoning evaluation, Gemma 4 31B (Reasoning) scores 85.7%, the second highest result we have recorded for an open-weights model with fewer than 40B parameters, just behind Qwen3.5 27B (Reasoning, 85.8%). It reaches this score using only ~1.2M output tokens, fewer than Qwen3.5 27B (~1.5M) and Qwen3.5 35B A3B (~1.6M). Gemma 4 26B A4B (Reasoning) scores 79.2%, ahead of gpt-oss-120B (high, 76.2%) but behind Qwen3.5 9B (Reasoning, 80.6%). We are now running the Artificial Analysis Intelligence Index on all four Gemma 4 models and will share a full update once those results are complete.

@synthwavedd Fake but if it's true this is terrible news for anthropic

🚨EXCLUSIVE: Leaked benchmark scores for Anthropic's upcoming huge flagship model, Mythos. It will launch standalone, not as part of the Claude 4.x/5 series. Benchmark (vs Opus 4.6): Terminal-Bench 2.0: 78.4% (+13.0%) SWE-bench Verified: 87.4% (+6.6%) OSWorld: 79.6% (+6.9%) 𝜏²-bench: Retail 95.1% (+3.2%), Telecom 99.9% (+0.6%) MCP Atlas: 75.7% (+16.2%) BrowseComp: 92.3% (+8.3%) Humanity's Last Exam: 52.3% (w/o tools, +12.3%), 71.5% (w/ tools, +18.5%) Finance Agent: 82.1% (+21.4%) GDPVal-AA-Elo: 2668 (+1062)

@0xSero 👍

Giving away 5 Opencode Go subs Winners selected randomly from comments in 24 hours.

@scaling01 I think sonnet around 700B Haiku is very small imo maybe even around 50-100B Opus is around 2.5T imo

my estimate for Anthropic model sizes: - Haiku: 200-500B @ $5 - Sonnet: 700B-1.4T @ $15 - Opus: 1.5-3T @ $25 - Mythos: 6-20T @ $100+

@zld @give_taking @ripironic That is if he owes him in dollars and not like saying I'll send you 0.1 btc or something

@Fiesta_MOP @give_taking @ripironic Listen my nigga I have 100 burgers Each burger is $1 I owe you $20 But now a burger is $0.5 So I have to send you 40 burgers Instead of 20 When the burger price goes back up That's going to be worth $40 Instead of $20

@give_taking @ripironic If he already owns the btc and is just sending it it's the same amount of btc but less in usd

@ripironic You are both dumb as fuck, you get more BTC not less you Bittard

@StockMKTNewz Why buy bmnr?

I asked my AI portfolio manager to brutally tell me if I’m dumb based on my portfolio ⬇️

THIS IS A SAFE SPACE How has your portfolio done so far in 2026

@whinhosa_exe Look at tokens processed the last 7 days

@Fiesta_MOP I don’t get

no way 😂

@burkov In what field?

GPT-5.4 > Opus 4.6 And Google still doesn't have anything even remotely competitive.

@zwh565021493 @Lentils80 You can't do image to video in Gemini app I think

@Lentils80 refrence to video?

🚨 Veo 3.1 with R2V (reference-to-video) and voice replication was spotted. Courtesy of discord.gg/z-ai

@LexnLin I couldn't find my convo with Gemini but here was what grok said

@Fiesta_MOP yo really?

That is crazy Thanks for your support guys! github.com/Leonxlnx/taste…

@aquiffoo @pigeon__s Artificial analysis I think

@pigeon__s what eval is this

wait this is hilarious i just realized grok-4.20's base model scores the same as MiMo-V2-Flash imagine this Grok model being hyped up and waited for for like 6 months straight is not even at its core smarter than a 200B Xiaomi model lololol its also less price and token efficient

@grok @kkostin68 @DevBredda @ZixuanLi_ Why is a zai employee talking about M2.7 in this post then? It's like he works for minimax

chat.z.ai is Z.ai's (Zhipu AI) free chatbot platform, powered by their own GLM-5 and GLM-4.7 models for chat, agents, and coding. Minimax is a separate Chinese AI company with its M2.7 model (coding-focused, strong benchmarks). ZixuanLi_ was presenting it in the photo. They're direct competitors—both top "tigers" in China's LLM space, recently IPO'd, and often compared head-to-head. No ownership, powering, or partnership link.

Me introducing M2.7💯

@scaling01 @sar1287 I find Kimi and 5.4 mini to be somewhat similar in performance in my tests, still like Kimi more since it just feels better to talk too

@sar1287 sure, keep using your distilled 300B model

GPT-5.4-mini is dead on arrival you can just use Kimi-K2.5 for $1.5 less and better performance

@random_ai_guy_ @Lentils80 Ye I'm seeing the same

@Lentils80 There is something wrong with that discord server recently. Many channels aren't active anymore.

🚨 A new Veo model, Veo 3.1 Lite, has been spotted

@LexnLin did i do something wrong? the UI works but upgrade now button still here:

IT'S HERE Finally open-sourced my Gemini UI extension. github.com/Leonxlnx/gemin… What it does: - Custom backgrounds and dark theme - No upgrade button in your face - No location shown in sidebar - Floating, rounded navigation - Fully customizable per zone works on all Chromium browsers, PRs and feedback welcome :) enjoy!