Sign in
Announcements we track

New releases

Models announced or first listed recently that are still waiting for an independent board we track — sorted newest first. Same corpus as the home “Newly tracked” strip.

Showing 1–24 of 257 · sorted by announced date ↓ · RSS
Nov 6
Pending benchmarksVideoAudioImage
V
HyperFlow H3 Flash
Video Rebirth · Video Rebirth's compressed few-step MiniMax-H3 derivative for joint video and synchronized stereo audio (not an official MiniMax product)
open weights
Nov 6
Pending benchmarksText
E
Manta
Empero · Empero's long-context model, announced with early access before a public release
API waitlist
Nov 5
Pending benchmarksTextImage
K
Kanana-2-v-decide
Kakao · Kakao's open-weight multimodal decision model (checkpoint kanana-2-v-decide-3b-2611, 3.67B parameters: 0.49B vision encoder + 3.18B language model) built on…
open weights
Nov 5
Pending benchmarksAudio
L
Accentus
Luel · Luel's model for detecting American English in real-world speech used for speech-data QA
Tracking
Nov 5
Pending benchmarksText
B
ERNIE-5.5-Flash-Base
Baidu · Baidu's open-weight pre-training checkpoint for ERNIE 5.5 Flash: a sparse MoE language model with 280B total parameters and 14B activated per token…
open weights
Nov 5
Pending benchmarksTextImage
F
GLiDE
Fastino · Fastino's open-weight GLiDE decision-model family: five sizes (0.4B, 0.8B, 2B, 9B, 27B) that share the System One schema (choice, noul yes/no, score) and…
API live · open weights
Nov 5
Pending benchmarksTextCode
O
GPT-6 Cyber
OpenAI · OpenAI's cybersecurity agent model for finding, validating and helping fix exploitable vulnerabilities while staying inside the scope researchers set
API waitlist
Nov 5
Pending benchmarksAudioVideoText
B
Higgs Realtime Duplex
Boson AI · Boson AI's research preview of the next generation of Higgs Realtime: a full-duplex interactive omni model that listens, watches and talks at the same time…
API waitlist
Nov 5
Pending benchmarksTextImage
T
Inkling 2 Small
thinky · Thinking Machines' second-generation small multimodal MoE (594B total parameters, 17B active) with native vision via a 292M Perception Encoder
API live
Nov 5
Pending benchmarksTextImageVideo
I
Ling-3.2-flash
Ant Ling (InclusionAI, Ant Group) · Ant Ling's native multimodal MoE model in the Ling series: 560B total parameters with about 25B activated per token, a 1M-token context window, and native…
API live
Nov 5
Pending benchmarksTextImage
H
THX-02
HAL-X · HAL-X's open-weight multimodal decision model: a 4.5B decision network plus a 60M think adapter that returns calibrated probabilities over options you supply…
API live · open weights
Nov 5
Pending benchmarksText
D
VINCO
Distributed Cognition · Distributed Cognition's specialist model for competitive 1v1 Catan
Tracking
Nov 4
Pending benchmarksText
P
PLaMo 3 Translate
Preferred Networks · Preferred Networks' open-weight translation-specialized LLM family (0.8B, 2B, 8B, 31B), built on PLaMo 3 NICT from joint work with Japan's NICT and further…
open weights
Nov 4
Pending benchmarksTextImage
O
Orvena 4B
Orvena · Orvena's open-weight on-device assistant model for the Orvena iPhone app: a LoRA fine-tune of Qwen3.5-4B (LoRA r=16, alpha 320 on all 248 linear layers…
open weights
Nov 4
Pending benchmarksText
M
IOTA Decide 4B
Macrocosmos · Macrocosmos's first public model: a 4.2B decision model fine-tuned from Qwen3.5-4B on GPUs spread over the internet, returning calibrated probabilities rather…
open weights
Nov 4
Pending benchmarksImageVideo
P
Pangram Image 1.0
Pangram Labs · Pangram's AI image and video detector, with a heatmap that localizes AI-generated regions; follows the Pangram Image research preview
API live
Nov 4
Pending benchmarksImageVideo
P
Pangram Image Flash 1.0
Pangram Labs · The faster, lower-cost sibling of Pangram Image 1.0 for detecting AI-generated images and video
API live
Nov 4
Pending benchmarksTextImage
L
Pyramid-JiT 2B (pjit-21k)
Linum · Linum's 2.2B pixel-space text-to-image research checkpoint trained for about $21K, with open weights and code
open weights
Nov 4
Pending benchmarksText
R
Ripple-65B Preview
Ricursive Intelligence · Ricursive Intelligence's dense 65B open-weight language model for agentic coding, pretrained from scratch on 12T tokens
API live · open weights
Nov 4
Pending benchmarksText
C
SWE-2 Flash
Cognition · Cognition's smaller coding model, mid-trained and RL-trained from Thinking Machines' Inkling-Small (276B total, 12B active parameters)
weights announced
Nov 4
Pending benchmarksText
W
TwIL-8B
webAI · webAI's 8.8B open-weight text model, post-trained from IBM Granite 4.2 8B with full supervised fine-tuning and GRPO/DAPO reinforcement learning
open weights
Nov 4
Pending benchmarksRobotics
V
Verne V0.5
Verne Robotics · Verne Robotics' open-weight world action model for industrial dual-arm mobile robots: a video transformer and an action transformer trained jointly, emitting…
open weights
Nov 4
Pending benchmarksVideo
V
Viggle Remake
Viggle AI · Viggle's video-to-video transformation model, released as open LoRA adapters for MiniMax-H3: recast characters, swap objects, change backgrounds and restyle…
open weights
Nov 3
Pending benchmarksRoboticsTextImage
L
Liber-0
LiberAI · LiberAI's World Action Model for zero-shot robot manipulation, pretrained on over 2.2M hours of human experience
Tracking