Skip to content
IFAI

AI models

New releases, benchmarks, pricing and API changes from GPT, Claude, Gemini, DeepSeek, Qwen and other large language models, updated daily.

11 stories in the last 7 days

  1. Reach 43

    Character.ai reveals early pretraining secrets, including 6-bit Squinch

    Character.ai just made public the internal engineering techniques that powered its early pretraining cluster, originally spearheaded by co-founder Noam Shazeer. The headliner is Squinch, a 6-bit gradient compression algorithm that matched bfloat16 fidelity despite operating on a fraction of modern network bandwidth. Why it matters: These bandwidth-saving tricks offer valuable architectural blueprints for anyone training sparse MoE models or fine-tuning across bandwidth-constrained clusters.

  2. Reach 54

    OpenAI builds automated red team to secure ChatGPT Atlas

    OpenAI just rolled out a major security update to ChatGPT Atlas's browser agent, powered by an RL-trained automated red team. The internal attacker proactively hunts for complex, long-horizon prompt injection exploits before bad actors find them in the wild, baking defensive updates directly into the model checkpoint. Why it matters: as agents get the keys to your inbox and accounts, staying ahead of prompt injections is the only way to keep them safe.

  3. Reach 49

    Google unveils T5Gemma 2 with multimodality and 128K context

    Google just launched T5Gemma 2, an encoder-decoder model family built on Gemma 3 that introduces multimodal vision and massive 128K token contexts. By tying embeddings and merging attention mechanisms, Google squeezed parameters down to compact sizes (~370M to ~7B) ideal for on-device deployment. It is available as pre-trained weights ready for downstream fine-tuning.

  4. Reach 51

    Google launches FunctionGemma for on-device tool use

    Google just released FunctionGemma, a lightweight model tuned from Gemma 3 270M specifically for on-device function calling. Small enough to run locally on phones and Jetson Nano boards, fine-tuning boosts its task accuracy from 58% to 85%. Why it matters: It turns edge devices into fast, fully private agents capable of triggering native actions without touching the cloud.

  5. Reach 53

    OpenAI launches GPT-5.2-Codex for heavy-duty coding and security

    OpenAI just rolled out GPT-5.2-Codex, an agentic coding model tuned for complex software engineering and defensive cybersecurity. It brings native context compaction for massive refactors, better Windows support, and sharper security auditing chops. It is available now in Codex for paid ChatGPT users, with API access rolling out soon.

  6. Reach 63

    Google's Gemini 3 Flash brings Pro brains at Flash speeds

    Google just expanded its flagship lineup with Gemini 3 Flash, delivering Pro-grade reasoning at triple the speed and a fraction of the cost of Gemini 2.5 Pro. It scored an impressive 78% on SWE-bench Verified—surpassing even Gemini 3 Pro—and is already rolling out globally as the default free model inside the Gemini app and Search.

  7. Reach 34

    Meta drops SAM Audio to segment sounds like images

    Meta just unveiled SAM Audio, bringing its "segment anything" vision to sound. The multimodal model lets users isolate specific sounds from messy audio mixtures using text prompts, visual clicks on video objects (like a guitar in a band), or time ranges. It's powered by a diffusion transformer architecture and available to test immediately in the Segment Anything Playground.

  8. Reach 51

    Google supercharges Gemini audio for live agents and earbuds translation

    Google just updated its Gemini 2.5 Flash Native Audio model to make real-time voice agents significantly better at multi-turn conversations and function calling. The tech also lands in the Google Translate app with a live speech-to-speech beta that streams translated audio straight into your headphones while keeping your natural tone and pitch.

  9. Reach 55

    OpenAI drops GPT-5.2 with expert-level knowledge work skills

    OpenAI just launched the GPT-5.2 model family—spanning Instant, Thinking, and Pro tiers—built specifically for professional workflows. The Thinking variant cuts hallucinations by 30% relative to GPT-5.1 and matches or beats human professionals on 70.9% of knowledge tasks across 44 occupations. It is rolling out to paid ChatGPT subscribers and API developers today.

  10. Reach 58

    Runway launches GWM-1, a real-time interactive world model

    Runway just announced GWM-1, its first family of general world models built on Gen-4.5 that generates frame-by-frame simulations in real time. It reacts to camera movements, audio, and robot actions across three specialized versions for explorable worlds, interactive avatars, and robotics. Why it matters: robotics teams can now test policies and generate counterfactual data inside a video diffusion simulator before touching physical hardware.

  11. Reach 45

    OpenAI's 'code red' response to Gemini 3 is landing next week

    OpenAI just pulled the emergency brake, with Sam Altman declaring a internal 'code red' to accelerate the release of GPT-5.2 to December 9th. The move is a direct counter to Google's Gemini 3, with internal benchmarks reportedly putting OpenAI back on top. While dates could still slip due to capacity, OpenAI is clearly feeling the heat.

  12. Reach 53

    Google's Gemini 3 Pro tackles spatial reasoning

    Google just introduced Gemini 3 Pro, upgrading its frontier model with advanced document, spatial, screen, and video reasoning. It can pinpoint pixel-precise coordinates, analyze video at 10 FPS, and reverse-engineer messy documents straight into structured LaTeX or HTML. Why it matters: Multimodal AI is shifting from basic image recognition to driving real-world robotics and desktop agents.

  13. Reach 53

    Google drops Gemini 3 Deep Think for heavy reasoning

    Google just rolled out Gemini 3 Deep Think to Google AI Ultra subscribers in the Gemini app. Aimed at tough math, science, and logic tasks, it uses parallel reasoning to test multiple hypotheses at once, hitting 41.0% tool-free on Humanity’s Last Exam and 45.1% on ARC-AGI-2 with code execution.

  14. Reach 52

    OpenAI is cooking up a new Garlic model to fight back Google

    OpenAI is reportedly fast-tracking a new frontier model codenamed "Garlic" for an early launch in the new year. Trained on a significantly smaller dataset than GPT-4.5 to keep costs down, it rivals Google's Gemini 3 and Anthropic's Opus 4.5 in coding and reasoning. With Google closing the user gap and Sam Altman declaring an internal code red, OpenAI is feeling the pressure.

  15. Reach 53

    DeepSeek drops V3.2 and reasoning-heavy Speciale models

    DeepSeek just launched DeepSeek-V3.2 and DeepSeek-V3.2-Speciale, two reasoning-first models built specifically for AI agents. The standard V3.2 is now live across its app, web interface, and API, while the higher-end Speciale reasoning model is currently limited to the API.

  16. Reach 70

    Anthropic launches Claude Opus 4.5 for coding and agents

    Anthropic just released Claude Opus 4.5, calling it the world's best model for coding, agents, and computer use. It even outperformed human candidates on the company's internal software engineering exam. With API pricing lowered to $5/$25 per million tokens and a new effort parameter to cut token usage, frontier intelligence is getting both smarter and cheaper.

  17. Reach 53

    OpenAI drops GPT-5.1-Codex-Max for marathon coding

    OpenAI just launched GPT-5.1-Codex-Max, an agentic coding model built to run continuously for over 24 hours. Using native context compaction, it works seamlessly across multiple windows on multi-million-token refactors while cutting thinking tokens by 30%. It is now the default engine inside Codex.

  18. Reach 56

    Google launches Gemini 3 with big reasoning leaps

    Google just unveiled Gemini 3, its next-generation frontier model family led by Gemini 3 Pro. The model claimed the top spot on LMArena with an unprecedented 1501 Elo, delivering huge upgrades in multimodal reasoning and vibe coding alongside a new agentic platform called Google Antigravity. Why it matters: Google is pushing hard to retake the AI crown by turning models into autonomous, self-executing software agents.

  19. Reach 57

    Claude lands inside Microsoft Foundry and Excel

    Anthropic just expanded its partnership with Microsoft, bringing Claude Sonnet 4.5, Haiku 4.5, and Opus 4.1 to Microsoft Foundry in public preview. Azure enterprise customers can deploy Claude through existing Azure agreements, while Excel's Agent Mode now lets users tap Claude to build spreadsheets and generate formulas. It removes procurement hurdles for teams wanting frontier reasoning inside Microsoft's ecosystem.

  20. Reach 51

    Elon Musk reveals Grok 5 with 6 trillion parameters

    Elon Musk just confirmed that xAI is actively training Grok 5, targeting an early 2026 release. Packing an eye-popping 6 trillion parameters—double the size of Grok 3 and Grok 4—the model will integrate real-time multimodal processing across text, vision, and audio, with Musk betting it will be the world's smartest AI by a wide margin.