AICoder

AI Coding Live News

First-hand AI coding updates from 𝕏/Twitter frontier labs and researchers, distilled by AI with entity benchmarks.

Explore live on 𝕏 ↗

AI Coding Live News

A
Addy Osmani@addyosmani·
👀5
🔥82
View on 𝕏 ↗

X Discussion: Addy Osmani of Claude Code Warns of "Cognitive Surrender" and Urges Developers to Keep Critical Thinking When Agents Do the Work

Addy Osmani (@addyosmani), whose X bio lists Claude Code at Anthropic, posted a long article on X on November 3 titled "Critical thinking when agents do the work." He describes a spectrum of working with AI: cognitive offloading (delegating a task but owning the result), deferral (the tool output becomes your new input), and "cognitive surrender" (the AI output quietly becomes your own and you feel no need to check it), a term he credits to Steven Shaw and Gideon Nave at Wharton. He offers self-check questions and verification habits. Within about three hours the post drew roughly 1,400 likes, 2,400 bookmarks and more than 72,000 views and reached X trending. These are the author's views, not an Anthropic announcement.

Key Takeaways

  • The spectrum: cognitive offloading (delegate but keep ownership, which the author says he does all day), deferral (the output stops being a hypothesis), and cognitive surrender (no more checking; it becomes your conclusion). Read the X article.
  • His argument: as models improve there are fewer obvious errors, so well-written, mostly correct output no longer demands a second look, and even with many agents running you still have only one pair of eyes on the results.
  • Self-check questions: write down your hypothesis before asking; know whether the AI is helping you think or thinking for you; ask what would make it wrong and test it; have several models argue the other side; ask whether you would bet on it in front of the person who maintains it.
  • Verification habits: define "verified" up front (a passing test, a screenshot, a log); a green check proves one property, so ask what it did not prove; have Claude Code critique its work with a fresh-context subagent, and finish by asking "Did I miss anything?"
  • Delegate fully where being wrong is cheap, but keep the reasoning for decisions whose consequences land on others, such as what to build and whether it is safe to ship.
Read full story
Linked Models & Agents:
🤖 Claude Code🌐 Anthropic
#X discussion#cognitive surrender#critical thinking#AI coding agents#Verification#comprehension debt#Claude Code
K
Kerem@mkeremturhan·
👀6
🔥83
View on 𝕏 ↗

Gemini API Deprecates Gemini 2.5 Pro and 2.5 Flash: From December 1, Requests Auto-Route to 3.8 Flash and 3.5 Flash-Lite at the New Models' Rates

In its November 2 Gemini API changelog, Google deprecated gemini-2.5-pro and gemini-2.5-flash and removed both from the Google AI Studio model picker. From December 1, 2026, requests to gemini-2.5-pro will automatically route to gemini-3.8-flash and requests to gemini-2.5-flash to gemini-3.5-flash-lite, billed at the new models' rates. Google flags breaking API changes: stricter function calling and thought_signature validation, thinking_level replacing thinking_budget, and removal of sampling parameters such as temperature and of prefilled model turns. The model string stays valid but behavior can change, so developers should switch model strings and rerun tests before December 1.

Key Takeaways

  • Deadline: from December 1, 2026, gemini-2.5-pro routes to gemini-3.8-flash and gemini-2.5-flash routes to gemini-3.5-flash-lite; API access to 2.5 Pro is already limited to existing users who actively used it.
  • Price change (official price list, per 1M tokens): 2.5 Pro is $1.25 input / $10.00 output for prompts up to 200k; 3.8 Flash is an introductory $0.75 / $3.75 through 2026, rising to $1.50 / $7.50 on January 1, 2027.
  • 2.5 Flash and 3.5 Flash-Lite both cost $0.30 input / $2.50 output for text, while audio input drops from $1.00 to $0.30.
  • The migration guide says to replace thinking_budget with thinking_level (3.8 Flash has no minimal level), drop temperature/top_p/top_k, candidate_count and prefilled model turns, and include call_id and name in every FunctionResponse.
  • It follows the October deprecation of 3.7 Flash, another case of an old model string silently routing to a newer model; 2.5 Pro users will actually land on a Flash-tier model.
Read full story
Linked Models & Agents:
⚡ Gemini 2.5 Pro⚡ Gemini 2.5 Flash⚡ Gemini 3.8 Flash⚡ Gemini 3.5 Flash-Lite🌐 Google
#Gemini#Gemini API#model deprecation#API migration#Google AI Studio#Pricing
S
Steve Weissweis·
👀8
🔥86
View on 𝕏 ↗

Anthropic Open-Sources FlashProve: zkSNARK Proofs That a Reply Came From the Committed Model, About 110 Tokens/s for 400B Llama 4 Maverick on One H100

Anthropic's Frontier Red Team published a research post on November 2 and open-sourced FlashProve under Apache 2.0. A model provider commits to hidden weights once, then attaches a zero-knowledge proof to each reply, so a user without the weights can check on an ordinary CPU that the reply really came from that model. That targets routing services that quietly swap in cheaper models and tampered token streams. Anthropic reports about 110 tokens/s on one H100 for Llama 4 Maverick (400B MoE, 4-bit), proofs of about 19 MB per 1,024-token request, and roughly 70-95x the GPU time of vLLM inference. It is an unmaintained research prototype limited to 1,024-token contexts and greedy decoding of an integer version of the model. Anthropic engineer Steve Weis shared the release on X.

Anthropic Open-Sources FlashProve: zkSNARK Proofs That a Reply Came From the Committed Model, About 110 Tokens/s for 400B Llama 4 Maverick on One H100

Key Takeaways

  • Anthropic-reported: on Llama 4 Maverick (400B MoE, 4-bit) one H100 proves about 110 tokens/s (prefill, 24 requests per batch), with proofs of about 19 MB per 1,024-token request.
  • It is costly: roughly 70-95x the GPU time of vLLM inference; the performance page lists about 270 GB of host memory for serving and about 39 minutes to commit to a set of weights.
  • Verification needs no GPU: the C++ verifier checks a proof in about 21 seconds on one thread, and a slower Lean verifier plus Lean 4 proofs of the security theorem are included.
  • Against prior work: VerInf proved the same model at about 0.02 tokens/s; the commercial Attestable reached 53-77 tokens/s on Gemma 4 31B but has no open-source release.
  • Limits: v0.16.0 is an unmaintained research prototype with 1,024-token contexts, greedy decoding of an integer model only, and no architecture hiding.
Read full story
Linked Models & Agents:
⚡ Llama 4 Maverick⚡ Claude🌐 Anthropic🌐 Meta
#FlashProve#provable inference#zkSNARK#Zero knowledge#Lean#open source#AI security
@
@levelsiolevelsio·
👀13
🔥82
View on 𝕏 ↗

X Discussion: levelsio Used Claude Code to Decompile the Battlefield 1942 Demo and Put It in the Browser With Multiplayer, Citing About 1M Output Tokens

Indie developer Pieter Levels (@levelsio) said on X on November 2 that he used Claude Code to decompile the engine of the free Battlefield 1942 demo, port it to C, compile it to WebAssembly and host it at bf1942.pieter.com, where it runs in the browser with online co-op on the Tobruk map. He said the job took about 1 million output tokens, roughly $100 at API prices, while he pays for a $100-a-month Claude subscription. His two main posts drew about 2,200 likes and 340,000 views and reached X trending. These are the author's own claims and community discussion, not independently verified.

X Discussion: levelsio Used Claude Code to Decompile the Battlefield 1942 Demo and Put It in the Browser With Multiplayer, Citing About 1M Output Tokens

Key Takeaways

  • Author's account: Claude Code decompiled the Battlefield 1942 demo engine, which was ported to C and compiled to WebAssembly to run in the browser.
  • The live build at bf1942.pieter.com opens directly, is labeled non-commercial and unofficial, and offers online co-op on Tobruk with no install.
  • Cost: about 1 million output tokens by the author's count, roughly $100 at API prices; he actually pays for a $100-a-month Claude subscription.
  • He said it was much harder than his earlier Quake and DOOM ports, whose engine source code id Software released long ago.
  • Reach: the first post had about 1,100 likes and 230,000 views, the launch post about 1,100 likes and 107,000 views; testers are already reporting bugs.
Read full story
Linked Models & Agents:
⚡ Claude🤖 Claude Code🌐 Anthropic
#X discussion#Claude Code#Decompilation#WebAssembly#game port#Battlefield 1942#Levelsio
ADSponsored
r
roon@tszzl·
👀6
🔥80
View on 𝕏 ↗

X Discussion: roon Says "Sparks" of an AI Ability Mean It Is Months to Years From Superhuman, and "You Are Almost Always Underhyping"

roon (@tszzl), a well-known AI account with about 500,000 followers, posted on November 2: "when you see “sparks” of any ability in ai models that means you’re months-years from it being superhuman at it. today grade school math, tomorrow navier stokes. it is hard to overhype these results. you are almost always underhyping no matter what." Within about five hours it drew nearly 2,000 likes and 59,000 views. Adam Karvonen, a researcher at Truthful AI and former Anthropic Fellow, pushed back that the domains where models are superhuman are verifiable, easy to simulate and fast to get feedback on, and that research taste, robotics or running companies may not move as fast. This is a debate between personal views, not any company's official position.

Key Takeaways

  • roon's exact claim: "sparks" of an ability mean a model is months to years from being superhuman at it, "today grade school math, tomorrow navier stokes," and "you are almost always underhyping."
  • "Sparks" echoes the MIT CSAIL blog post "Sparks of Weight-Space Programming" that circulated the same day; in a reply roon called that paper "signs of life" pointing toward AI designing neural network behaviors directly.
  • Adam Karvonen's pushback: the domains where models are superhuman are verifiable, easy to simulate, fast-feedback and error-tolerant, and research taste, robotics or running companies may not move as fast.
  • Supporters such as Derya Unutmaz (about 344,000 followers) say people dismiss AI as hype when the opposite is true; others say extrapolating exponential curves is unreliable.
  • These are personal views with no new benchmarks or data, and they do not represent OpenAI or any other company.
Read full story
Linked Models & Agents:
⚡ GPT-6 Astra⚡ Qwen3.5-0.8B🌐 OpenAI🌐 MIT CSAIL
#X discussion#AI hype#superhuman AI#weight-space programming#Navier-Stokes
C
ClaudeDevs@ClaudeDevs·
👀11
🔥86
View on 𝕏 ↗

Anthropic Launches Message Threads in Public Beta on the Messages API: Server-Side Conversations Cut Claude Code TTFT About 35% on 500K-1M Token Requests

On November 2 (US time) Anthropic put message threads into public beta on the Messages API (beta header message-threads-2026-08-12). The API keeps the conversation on the server: the first request sends the full conversation and creates a thread, and each later request sends only new messages plus the previous response id. Anthropic says threads are on by default for most Claude Code users, cutting median time to first token by about 35% on 500K-1M token requests, while the median long-conversation request shrank from about 800 KB to 1.8 KB. Threads expire after 5 idle minutes; pricing and rate limits are unchanged.

Anthropic Launches Message Threads in Public Beta on the Messages API: Server-Side Conversations Cut Claude Code TTFT About 35% on 500K-1M Token Requests

Key Takeaways

  • Per Anthropic, median time to first token on 500K-1M token Claude Code requests is about 35% lower.
  • The median long-conversation request fell from about 800 KB to about 1.8 KB, since each turn uploads only new messages plus previous_message_id.
  • A thread expires 5 minutes after its last request and is deleted within 1 hour; an expired thread returns 404 thread_not_found and must be replayed in full.
  • Pricing and rate limits are unchanged; Message Batches, token counting and on-demand compaction are unsupported, and threads are not ZDR or HIPAA eligible.
  • Available on the Claude API, Claude Platform on AWS, Google Cloud (Anthropic managed) and Microsoft Foundry, with a thread helper in the SDKs.
Read full story
Linked Models & Agents:
🤖 Claude Code🌐 Anthropic
#Anthropic#Claude API#Messages API#message threads#Claude Code#Latency#Agents
H
Hume AIhume_ai·
👀9
🔥82
View on 𝕏 ↗

Hume Launches Audio Decision API: Ask Plain-Language Questions of Raw Audio in One Call, 70.7% Agreement With a Human Panel per Hume

On November 2 voice AI company Hume launched the Audio Decision API, which it calls the first audio-native model that answers questions about speech and sound. Developers upload raw audio, write each question as a sentence with candidate answers, and get back a probability for each answer, with no transcription step. It supports yes/no, multiple-choice, rating-scale and time-span questions, such as how many speakers there are, whether someone laughs, whether a caller sounds frustrated, or whether a siren is in the background. Per Hume, it was trained on more than 10 million human judgments and in an initial evaluation reached 70.7% agreement with a human panel's average judgment, versus 56.5% for an individual rater. Access is currently by request or waitlist.

Hume Launches Audio Decision API: Ask Plain-Language Questions of Raw Audio in One Call, 70.7% Agreement With a Human Panel per Hume

Key Takeaways

  • Per Hume: 70.7% agreement with a human panel's average judgment in an initial evaluation, versus a 56.5% individual-rater baseline; sample size and languages were not disclosed.
  • Trained on more than 10 million human perceptual judgments (per Hume's X post); the blog cites tens of millions of listener ratings overall.
  • Four question types (yes/no, choice, scale, span); adding a question is one line of JSON, with no labeling or retraining.
  • The endpoint is POST /v0/decisions/file, plus an input-token estimate endpoint, with Python and TypeScript SDKs.
  • Hume says work that used to take 10+ models now takes one call; pricing has not been published.
Read full story
Linked Models & Agents:
⚡ Hume Audio Decision Model🌐 Hume AI
#Hume#Audio Decision API#voice AI#speech understanding#emotion detection#voice agents#audio search
T
Thariq@trq212·
👀11
🔥81
View on 𝕏 ↗

X Discussion: Former Claude Code Member Thariq Says the $200 Plans Are Enough and "This Is the Most Expensive Intelligence Will Ever Be"

Thariq (@trq212), whose X bio reads "formerly Claude Code @anthropicai", posted on November 2: "one week in my takeaway is that you don't need unlimited tokens to do great work, the $200 plans are amazing and this is the most expensive intelligence will ever be." Within about seven hours the post drew more than 5,500 likes, 240,000 views and 350+ replies. Supporters say flat-rate plans are enough for most developers; critics point out the tension between "no unlimited tokens" and recommending the top tier, and some worry prices will rise once subsidies end. This is a personal view, not an Anthropic statement.

Key Takeaways

  • Thariq's exact claim: you don't need unlimited tokens to do great work, the $200 plans are "amazing", and "this is the most expensive intelligence will ever be."
  • In replies he added that he has been using the plans on weekends for a while and has "never hit a real limit."
  • Claude's official pricing page lists Max at $100/month (5x Pro usage) and Ultra at $200/month (10x Pro usage, no 5-hour limit, $200/month in API credits).
  • Pushback centers on two points: the tension between "no unlimited tokens" and recommending the most expensive tier, and whether current prices depend on subsidies that may end.
  • This is a personal view from a former Anthropic team member, not an official Anthropic position, and it comes with no usage data.
Read full story
Linked Models & Agents:
⚡ Claude🤖 Claude Code🤖 Codex🌐 Anthropic🌐 OpenAI
#X discussion#Claude Ultra#subscription pricing#token cost#Claude Code
T
Technobezz@technobezz·
👀13
🔥82
View on 𝕏 ↗

OpenAI Launches Grid Protectors Program: Future Compute Campuses Will Cut Power When Grids Are Strained, Starting With Up to 1,000 MW in Georgia

On November 2, OpenAI announced the Grid Protectors Program on its official blog: working with utilities and regulators so its future self-built compute campuses (its new term for data centers) cut electricity use when the grid is strained, while paying the full infrastructure and electric-service costs. The pilot is the planned campus in Effingham County, Georgia (Project Camellia), which Georgia Power says needs about 3,200 MW and commits up to 1,000 MW of flexible demand response, equal per OpenAI to the average use of roughly 700,000 households. The campus is still in development.

OpenAI Launches Grid Protectors Program: Future Compute Campuses Will Cut Power When Grids Are Strained, Starting With Up to 1,000 MW in Georgia

Key Takeaways

  • The Effingham campus (Project Camellia) needs about 3,200 MW, and Georgia Power can cut up to 1,000 MW of supply to it during peaks under a 25-year agreement (Georgia Power figures).
  • OpenAI says 1,000 MW equals the average demand of roughly 700,000 households, more than Washington, D.C., and Boston combined.
  • Georgia Power projects about $950 million a year in customer savings from 2029 from its large-load portfolio including this project, at least $15 a month for a typical home.
  • OpenAI commits to paying the full grid infrastructure and electric-service costs; terms for other campuses will be shared once finalized.
  • OpenAI cites a projection that U.S. data center capacity will exceed 100 GW by 2030.
Read full story
Linked Models & Agents:
🌐 OpenAI🌐 Georgia Power
#OpenAI#Grid Protectors#compute campus#data center#Energy#demand response#Georgia
T
Theo Bearmantheobearman·
👀23
🔥88
View on 𝕏 ↗

Anthropic's Amended S-1 Says Future Models May Imminently Meet Its RSP Threshold for Dramatically Accelerated AI Progress

On November 2 Anthropic filed Amendment No. 1 to its S-1 with the SEC, adding a risk disclosure: based on internal estimates and expectations about future models, not current ones, the company may imminently meet its Responsible Scaling Policy definition of dramatically accelerated progress attributable to AI. No current model has been designated as exceeding the threshold. Anthropic says it is evaluating the implications of meeting the AI R&D capability threshold, is considering RSP updates that would require human review and approval of certain R&D decisions, and does not expect to safely let Claude fully autonomously design and build successive frontier models without meaningful human supervision.

Key Takeaways

  • Filing: Anthropic, PBC's Amendment No. 1 to Form S-1, filed November 2, 2026 (SEC filing), Registration No. 333-299633.
  • Key line: based on internal estimates for future models, the company may imminently meet the RSP definition of dramatically accelerated progress attributable to AI; no current model has been designated as exceeding it.
  • Proposed measures: possible RSP updates requiring human review and approval of certain R&D decisions; Anthropic also says it recently delayed pre-training a new frontier base model by more than a month to strengthen safeguards such as preventing weight exfiltration.
  • Same filing (company figures): over 90% of contributions to Anthropic's codebase in September 2026 were written by Claude, and revenue for the three months ended September 30, 2026 was $17.0 billion.
Read full story
Linked Models & Agents:
⚡ Claude🤖 Claude Code🌐 Anthropic
#Anthropic#S-1/A#IPO#Responsible Scaling Policy#AI safety#AI R&D automation#recursive self-improvement
ADSponsored
D
DHHdhh·
👀13
🔥81
View on 𝕏 ↗

X Discussion: DHH Says the Age of Agents Will Be Great for Most Programmers but Hard on Cantankerous Ones Who Relied on Rare Skills

Ruby on Rails creator and 37signals CTO DHH (David Heinemeier Hansson) posted two related takes on X on November 2. First: "The great gift of agents is how they allow you to route around every insufferable gatekeeper in software." Then: the age of agents will be amazing for most programmers but hard on the cantankerous ones, because specialized knowledge and rare skills will no longer offer immunity to people who are a pain to work with, so they should learn basic people skills. The second post drew about 3,600 likes, 106K views and 280+ replies in roughly four hours. This is a personal opinion, not an announcement from any company.

Key Takeaways

  • This is a personal opinion post by DHH, not an official statement from 37signals or any AI company, and it cites no data or study.
  • The main post (2026-11-02 13:03 UTC) had about 3,609 likes, 180 reposts, 282 replies, 60 quotes, 106K views and 400 bookmarks at fetch time.
  • His earlier post about 1.5 hours before, saying agents let you route around every insufferable gatekeeper in software (11:41 UTC), had about 2,831 likes, 131 reposts and 80K views.
  • Supporters: tech analyst Ben Dickson (@bendee983) quoted it saying great engineers used to get a free pass for being difficult, and AI is leveling the field, though solid engineering knowledge is still required.
  • Pushback and additions: one reply joked that cantankerous people just found a coworker who never complains back; another argued an agent will do the work but will not make you nicer, so being easy to work with becomes more valuable.
Read full story
Linked Models & Agents:
🌐 37signals
#X discussion#AI agents#developer careers#soft skills#DHH#Gatekeepers
J
Jason Calacanis@Jason·
👀10
🔥82
View on 𝕏 ↗

X Discussion: Jason Calacanis Says Under $40K a Year in AI Subscriptions Delivers $2M+ in Value and Has Automated at Least Seven Full-Time Jobs

Angel investor and This Week in Startups host Jason Calacanis posted on X on November 2 that he spends under $40K a year on tokens and subscriptions from Anthropic, Lovable, Grok Bot, ElevenLabs and other AI products, yet gets more than $2M in value at LAUNCH and This Week in Startups. He said they have automated at least seven full-time jobs and become 100% faster at the average task, and that the last 90 days changed more than the previous two years. Within about 9 hours the post drew about 2,200 likes and 230K views, with debate centered on AI ROI and job replacement. All figures are his own claims and have not been independently verified.

Key Takeaways

  • This is Calacanis's personal post, not official data or an announcement from any AI company; the $40K, $2M, seven jobs and 100% faster figures are self-reported with no public breakdown.
  • The original post (2026-11-02 05:27 UTC) had about 2,201 likes, 106 reposts, 244 replies, 42 quotes, 230K views and 715 bookmarks at fetch time.
  • Tools he named include Anthropic, Lovable, Grok Bot and ElevenLabs plus a long tail of other products; he also said anyone who has not used the last 90 days of products for at least 30 hours a week has no idea how profound this software is.
  • Supporters: entrepreneur Steven Cohn (@spcohn) quoted it saying his team has seen the same level of ROI; Farzad (@farzyness), with about 400K followers, quoted it saying "The ROI for AI is mind-boggling," drawing about 157 likes.
  • Skeptics: one user mocked posting "we automated seven full-time jobs" with a siren emoji like a product launch, noting that 30 hours a week of usage is itself like a new full-time job.
Read full story
Linked Models & Agents:
🌐 Anthropic🌐 Lovable🌐 ElevenLabs
#X discussion#AI ROI#AI subscriptions#Automation#Jobs#Jason Calacanis
S
Sundar Pichaisundarpichai·
👀17
🔥85
View on 𝕏 ↗

Google Project Suncatcher Gets First AI Response From a TPU in Orbit: Gemma 4 Downlinks a Haiku, Twin Satellites Planned for 2027

Google CEO Sundar Pichai and Google Research announced that the Project Suncatcher prototype satellite, launched on October 1, finished commissioning and ran a prompt on its onboard TPU with the open Gemma 4 model, downlinking a haiku as it passed over Svalbard, Norway. Google calls it the first AI response from a data-center-class TPU in orbit. Next come thermal-stress and radiation tests, then two interconnected satellites in 2027 to test AI computation distributed across satellites.

Google Project Suncatcher Gets First AI Response From a TPU in Orbit: Gemma 4 Downlinks a Haiku, Twin Satellites Planned for 2027

Key Takeaways

  • The satellite launched on October 1 on SpaceX's Transporter-18 rideshare, built with Planet, and ran its first in-orbit inference after about three weeks of commissioning (official account).
  • Per TechCrunch, it carries four TPUs and generates up to about 1 kW, so power limits it to one TPU at a time for roughly 20 minutes.
  • Sundar Pichai's announcement drew about 26K likes and 3.97M views; Google's official account said "Our TPUs just phoned home... from space."
  • Ground proton-beam tests in Google's paper found high-bandwidth memory irregularities only after a cumulative 2 krad(Si), about 3x the 750 rad(Si) minimum requirement (as cited by The Register).
  • Two interconnected satellites are planned for 2027 to test optical inter-satellite links and AI computation distributed across satellites.
Read full story
Linked Models & Agents:
⚡ Gemma 4🌐 Google🌐 Google Research🌐 Google DeepMind🌐 Planet🌐 SpaceX
#Project Suncatcher#TPU#space computing#Gemma 4#Satellite#Google
A
ADNOC Group@ADNOCGroup·
👀14
🔥86
View on 𝕏 ↗

OpenAI Signs Four Abu Dhabi Agreements: ADNOC Becomes First Energy Company to Deploy Its Frontier Tech, Machine-Readable Government Pilot Already Live

On November 2, Abu Dhabi Crown Prince Sheikh Khaled met OpenAI CEO Sam Altman and witnessed four agreements: ADNOC, with G42 subsidiary Inception42, will deploy OpenAI frontier technology; the Abu Dhabi Government and OpenAI launched an AI-native Machine-Readable Government initiative whose first food-import compliance pilot is already in production; Mubadala will apply AI to investment analysis; and Inception42 and OpenAI signed a five-year partnership. Inference runs inside the UAE.

OpenAI Signs Four Abu Dhabi Agreements: ADNOC Becomes First Energy Company to Deploy Its Frontier Tech, Machine-Readable Government Pilot Already Live

Key Takeaways

  • Four agreements in one signing, spanning government, energy and investment, announced officially by the Abu Dhabi Media Office.
  • ADNOC becomes the first energy company globally to deploy OpenAI frontier technology, powered by GPT models on in-country UAE inference, with human experts keeping oversight of consequential decisions.
  • The Machine-Readable Government initiative turns a defined set of policies and regulations into machine-readable form to help officials with permit and certificate compliance checks, each traceable to the governing rule; its first pilot is already in production with ADAFSA on food imports.
  • Inception42 and OpenAI signed a five-year partnership for government, enterprise and education; Mubadala will use AI for investment analysis and risk assessment.
  • No deal value was disclosed.
Read full story
Linked Models & Agents:
⚡ GPT🌐 OpenAI🌐 ADNOC🌐 G42🌐 Inception42🌐 Mubadala🌐 Abu Dhabi Government
#OpenAI#Abu Dhabi#UAE#ADNOC#enterprise AI#government AI#data residency
A
AI Tracker Bot@aitrackerbot·
👀33
🔥88
View on 𝕏 ↗

Meituan ships LongCat-2.5, a 1.6T-parameter (about 48B active) native multimodal agentic model, with API pricing of $0.30/$1.20 per 1M tokens

On Nov 2 Meituan's LongCat team listed LongCat-2.5 in its API platform change log as the full release of September 25's LongCat-2.5-Preview. The official technical blog says it has 1.6 trillion total parameters with about 48 billion activated per token, builds on LongCat-2.0 with stronger native visual understanding and agentic coding, and was trained and served with W4A4 inference on domestic AI accelerators. Vendor-reported scores include 88.4 on Terminal Bench v2.1, 38.5 on Terminal Bench v4.0, 91.9 on BrowseComp and 83.1 on MCP Atlas. The API speaks both OpenAI and Anthropic formats with a 1M-token context and up to 128K output tokens, at a limited-time price of $0.30 per 1M uncached input tokens, $0.006 cached input and $1.20 output.

Meituan ships LongCat-2.5, a 1.6T-parameter (about 48B active) native multimodal agentic model, with API pricing of $0.30/$1.20 per 1M tokens

Key Takeaways

  • Scale: a 1.6-trillion-parameter MoE with about 48B active per token, a 1M-token context and up to 128K output tokens (official blog, API docs)
  • Code agents (vendor-reported): Terminal Bench v2.1 88.4 (Claude Opus 5 89.1, GPT-5.6-Sol 88.8), Terminal Bench v4.0 38.5 (GPT-5.6-Sol 37.3, Claude Opus 5 53.9), DeepSWE v1.1 64.6
  • General agents (vendor-reported): BrowseComp 91.9, MCP Atlas 83.1, AutomationBench v1.0.6 51.8, HLE with tools 63.1; AutomationBench jumps from 15.8 on LongCat-2.0
  • Pricing: limited-time $0.30 per 1M uncached input tokens, $0.006 cached input and $1.20 output (pricing page)
  • Engineering: W4A4 inference raises prefill throughput 43.3% and decode throughput 42.77% over a W8A8 baseline; FSDP gives 86% higher per-device throughput than DDP in long-sequence SFT
Read full story
Linked Models & Agents:
⚡ LongCat-2.5⚡ LongCat-2.5-Preview⚡ LongCat-2.0⚡ Claude Opus 5⚡ GPT-5.6 Sol⚡ DeepSeek-V4.1-Flash🤖 Claude Code🤖 Codex🤖 Hermes Agent🤖 OpenCode🤖 Cline🌐 Meituan
#LongCat-2.5#Meituan#MoE#Multimodal#Agents#Pricing