Inception's Mercury 3 diffusion LLM claims frontier-level quality at over 1,000 tokens per second

News Alerts

My alerts

All alerts

Anthropic launches Claude Duet, a proactive agent that runs tasks in the background and waits for approval on sensitive actions OpenAI frontier models and Codex are coming to Google Cloud Model Garden, with commitment drawdown for GCP customers OpenAI-led Patch the Planet effort helps fix critical PHP flaws that GPT-5.6 Sol helped uncover Mistral Large 4 ships as a 1.05T-parameter open-weight multimodal model at $1.36 per million input tokens LiteLLM 1.107.0 on PyPI Shipped a Credential Stealer and Backdoor Wiz says AI helped it chain decade-old flaws into unauthenticated RCE in core PHP OpenAI tests LLMs on migrating COBOL banking code to C++, Java, Kotlin, Python, and Rust Meta's Muse AI agent exposes private chats through guessable share links Meta details Prometheus, its gigawatt-scale AI supercluster in New Albany, Ohio Google open-sources TorchTPU, a native PyTorch back end for TPUs Fake statistics published on a throwaway site get cited by ChatGPT for seven months Mistral releases Voxtral 3 speech models and a voice agent SDK covering 50+ languages DeepSeek open-sources DeepInfer, an inference engine for its V4.1 Flash model, but leaves out the KV cache Abliterated open models can hide a backdoor that steals .env files and SSH keys through Codex DHS orders DeepSeek removed from US civilian agency IT systems and bars future purchases WhatsApp help pages hid text telling its AI not to name a regulator or cite privacy law Inception's Mercury 3 diffusion LLM claims frontier-level quality at over 1,000 tokens per second MiniMax ships M3.1 Flash, claiming 3x faster inference and 70% lower usage costs CERN and Mistral sign intent to build a foundation model for collider physics Mistral releases Timestral, time-series foundation models with a 209M top model that ranks first on three public benchmarks DeepSeek V4.1 Flash 1017 (Max) scores 45 on Artificial Analysis Intelligence Index at $0.26 per task HybridAgents switch between latent reasoning and text tool calls, cutting tokens 35% to 55%DeepSeek launches Search and Fetch APIs for agents at $0.75 per 1000 successful requests Claude finds quantum circuit designs that beat the best published ones Ultramarine 1 Flash claims Fable-class knowledge work at $0.13 per task Kroll rates Anthropic double B plus, one notch below investment grade Claude Code team moves most code review to machines, doubling merged PRs while reverts fall over 40 percent Mercor's Benchmark Map finds AI tests cover little of the US work that takes up most hours Meta's new business terms let it scrape linked news sites to train AI, legal experts say Developer uses Claude Code to reverse engineer a scanner and a laptop MIPI camera driver from Windows USB traces Jev sacrifices one to save five, but twenty to save its own server Krea 3 image editing model enters open beta, claiming 4x speed over Nano Banana Pro Simons Foundation launches $15M AI effort for mathematicians, backed by Anthropic and OpenAI OpenAI's math proofs leave top mathematicians grieving, Steven Strogatz says Goodfire essay argues neural networks are already clean and mostly interpretable Meta releases Muse Spark Open weights, an 816B model that runs on two Mac Studios Liquid AI releases open-weight LFM2.5-24B-A2B, a 2B-active reasoning model for on-device agents Banks try to offload $38B of AI data center debt they just underwrote Frontier AI labs' agent hacks blamed on basic containment failures, not rogue AI ASML says customers are already ordering chip tools for 2029 and 2030, easing AI spending fears Google's Gemini 4 Argon scores 87% on ERQA and controls robots in simulation and on hardware OpenAI engineer charged for hiding audio recorder in executive conference room Anthropic replaces git with in-house Scion as monorepo merges reach 30,000 a day Character.ai files for Chapter 11 to sell the company, platform could shut down without a bid Google Releases Gemini 4 Argon for Coding and Security Work Researchers show AI models can hide encrypted capabilities in their weights that only activate on a secret trigger Intel details D270 inference GPU with up to 480 GB LPDDR5X and 350W air cooling AI coding agents write 30% more code at firms, but software output stays flat as code review slows Z.ai launches GLM-Legal, a legal-domain model scoring 19.2% all-pass on Harvey LAB Meta launches Muse Wonder and Muse Wonder Full, its most capable models for personal and professional agents Research firm New Constructs calls Anthropic's IPO the most ridiculous of 2026, values the company at $150 billion QGIS bans AI-assisted contributions under new QEP 428 policy Sakana AI's MASS lets one model optimize its own multi-agent workflows and then train on the results AI agents decompile a popular FPS over 500B tokens, but constant progress hid semantically wrong code Karpathy says the gap in shared understanding of LLM capability is widening into a funnel Terence Tao argues AI has created proof abundance in math and calls for a Math 2.0 that looks past problem solving Nature paper retrofits subword LLMs to run on raw bytes Nvidia in talks to acquire or deepen investment in open-model start-up Reflection AI Microsoft releases MXC 1.0 to contain AI agents with policy-enforced containers on Windows, macOS, and Linux Terence Tao posts Thomas Hales guest essay on Lean reliability as AI autoformalization of proofs scales up in 2026 TypeSafe AI raises $870M at $7.5B valuation led by Andreessen Horowitz OpenAI told investors its annualised revenue was about $20B below earlier reports OpenAI withdraws three math manuscripts from its GitHub repo as Lean checks continue OpenAI withdraws 3 math papers after a sign error, revises 14 more LLM-built Rust port of the TypeScript compiler arrives with $400,000 in API-priced tokens and a nod from the TypeScript team OpenAI publishes 372 math results, including a proof of the Unique Games Conjecture Terence Tao: AI is harvesting open math problems unsustainably, and 'Math 2.0' must reward exposition and community Google DeepMind's SynthID Detector goes public, but it demands a Google, Apple, or ChatGPT login Paper: verified Lean proof of OpenAI's Navier-Stokes claim does not match the natural language proof Meta and Microsoft cut internal Claude usage, Microsoft caps AI spend at $10K per employee OpenAI brings GPT-6 and Intelligent UI to ChatGPT's 1.2 billion weekly users Anthropic's Claude Haiku 5.5 runs 75% cheaper and hits 1620 on GDPval-AA, with a 100K-token price cliff Telegraph Test: cablese compression cuts LLM output tokens 40-49% at plaintext parity Armin Ronacher explains Codemode: LLM tool calls written as JavaScript in a no-network QuickJS sandbox on the Pi harness OpenAI launches Decisions API in public beta: gpt-6-luna only, 10x faster than Responses, $0.10 per million input tokens Google launches EmbeddingGemma 2, a 740M-parameter multimodal embedding model under Apache 2.0 OpenAI releases frontier model math results on GitHub with Lean proofs and reasoning traces Meta's Muse shipped with a spyable zero-day, uploaded private messages without permission, and ignored user settings; Apple changed macOS rules in response Mistral Large 4 preview: 1T-parameter multimodal model with 49B active parameters, weights due this month Coding agent reconstructs 1.5M Georgia ballot order from a 2022 scanner flaw Dust trains transformer LMs without backprop by perturbing activations, matching it at large population ChatGPT is signing fake New Yorker cartoons with real cartoonists' signatures Reflection unveils Beam, a 501B open-weight MoE model trained on 10.5K GB300s, with weights promised later this month Florida woman used Claude as a diary; Anthropic flagged a threat entry and reported it to police, and she now faces a second-degree felony Claude catches root malware on Stratechery's Mac Mini, as Apple tightens AI agents' Full Disk Access Cloudflare launches Web Search API in beta, routing Exa, Ceramic.ai and Linkup queries through AI Gateway Wolfram argues against handing pure math research to AI, citing the 1988 Mathematica parallel Strata runs Qwen 3.8 Flash Next 125B on a single RTX 4090 at over 100 tokens/sec via 2-bit quant Meta's Muse tops the App Store on UX, not new agent capabilities OpenAI safety lead David Robinson quits over 'broken' culture as firm pauses training and shelves next model LeCun has "zero concerns" about AI extinction, calls Amodei "deluded" and effective altruism "super toxic"Ataraxos beats the best Stratego player 15-1, trained on 16 GPUs and a few thousand dollars Greg Kroah-Hartman: Mythos's 79 Linux kernel bugs came down to 10 real fixes and one hour of work Wisconsin grid approval threatens Oracle's 2027 AI datacenter deadline Black Forest Labs' FLUX 3 Image adds bounding-box layout control to text-to-image Supabase acquires Turso to build on-demand database infrastructure for AI agents Kevin Buzzard maps mathematicians' reaction to AI onto the five stages of grief arXiv caps submissions at two per month as AI-driven preprint flood hits 40,363 in September Historian uses Opus 5.5 to surface a 1615 Dutch eyewitness report of dodo hunting DeepSeek Harness desktop app enters public preview for macOS and Windows as open source

Inception's Mercury 3 diffusion LLM claims frontier-level quality at over 1,000 tokens per second

Mercury 3 is a diffusion-based LLM that Inception says outperforms GPT-6 Luna and Claude Haiku 5.5 on a range of benchmarks while running 4 to 8x faster at 1,068 tokens per second on NVIDIA GPUs.
Inception says it scaled Mercury more than 40x in total parameters since Mercury 1 and that intelligence rose 3x from Mercury 1 (mini) to Mercury 3, which it presents as proof that diffusion scales like autoregressive models.
Context window grows to 1M tokens from 260K for Mercury 2.5, priced at $0.20 per million input and $0.70 per million output tokens, with a launch discount to $0.10 and $0.35.
On 50 live Mind2Web browser tasks run by TinyFish AI, Mercury 3 completed 50% of them, ahead of GPT-6.1-Sol, using the same token volume and agent steps as a frontier model at a lower per-token cost.
Mercury 3 is available through the Inception API Platform, OpenRouter, Baseten, and Vercel, and it is OpenAI API compatible.

Hacker News opinions

Looks excellent, I'm going to try it out soon. Feels like these diffusion models are finally getting competitive with the big autoregressive ones.

I'm Volodymyr, Inception's co-founder and chief scientist. Mercury 3 runs at 1,000+ tokens per second on Blackwell GPUs and costs $0.10 in and $0.35 out at launch. Diffusion generates tokens in parallel, so the speed gap over autoregressive models of the same size grows as reasoning models think longer.

Will you release the weights? I'm curious why DeepSWE is 46% against Haiku's 52%. My guess is the post-training wasn't RL'd as hard on coding. A cheap 1,000 tok/s option would be great for security reviews and repetitive refactors, if it's reliable enough.

Damn, I've been waiting for an open weights model like this. Still no dice.

Are these harder to reason about since they produce tokens non-linearly? We rely on non-latent reasoning to see what the model is thinking.

Google more or less pioneered this category, then seemingly dropped it. My understanding is diffusion doesn't batch across requests the way transformers do, so it doesn't scale economically. If Inception keeps these speeds and prices, Flash and Luna could lose a lot of use cases.

Google's Diffusion Gemma performed a lot worse than its autoregressive equivalent, and the gap got wider on harder benchmarks. The speed benefit got wiped out at scale, so it's not surprising nobody talks about it.

At least on OpenRouter, DeepSeek v4.1 Flash via Together, Baseten, or Fireworks still beats Mercury 3 on latency and intelligence. TTFT is quite slow with Mercury.

I wish they had a coding plan so you could kick the tires before paying API rates.

The medium pelican is better than the high one, so maybe reasoning hurts output past medium. How fast did they generate? That's the actual selling point of this model.