SkillPedia Skills Encyclopedia
中文 EN
Today's updates 2026-11-17 UTC

No new commits yet today

Open the update feed

Models

Every model, and where it comes from.AI models by country and their current versions

The skill index answers “what skills are there.” This page answers a different question:which models exist right now, at what versions, from whom. Grouped by country and region, with domestic models first.

191 models 19 countries and regions 97 organizations 89 with open weights 8 types 146 with an API

Read the caveats firstRead the caveats first

58 domestic and 133 international — and the international group is mostly the United States, with most other countries contributing only one or two. So this list is not a comparison of national AI strength. It is an index organized by origin, answering “which models does a given country have right now, and at what versions.”

Versiontakes the latest available version from each vendor's site and release notes at the time of research. A single family often ships several tiers at once — Pro / Flash / Preview — and the version column lists them all rather than picking one and pretending it is the whole story.

Priceis the official list price at the time of research, quoted in USD per million tokens (input / output). Open-weight models you can deploy yourself are marked “Free / self-hosted”; models with no published pricing are marked “Not disclosed” — those two mean completely different things, so they are not merged into a single dash.

147 of 191 have a confirmed release date; the rest are left blank — nothing isinferred; every count and share on the page is computed from the database in real time, with no hardcoded numbers.

Abilitytakes the explicit wording from each vendor's docs and public material: confirmed support gets a solid chip, confirmed lack of support gets a dashed one, and anything not found during research isleft out entirely — an empty slot and “not supported” are two different things, and they are not merged.

Released179 Preview10 Unconfirmed2

Release status is the fourth axis, added 2026-10-21. The badge under each model name answers "can I use it today" — Released means GA and on the API; Preview means Preview/Beta or access behind an application or verification; In development means officially confirmed but not open yet; Unconfirmed means leaked or spotted in blind tests with no official announcement.

Decision models are a new kind of entry (8 so far): they generate no text at all and only return typed, probability-weighted verdicts for a given input (block or allow, route to which model, hand off to a human). So the context window, max output and function calling shown on these cards are “not applicable” rather than “not found” — printed as “—” instead of being padded out with a plausible-looking number.

Ability notes, usage fees, max output, response speed, rate limits and data-policy notes are Chinese-only for now, so they are left out on this English page. The Chinese version of each card shows them in full. Raw parameters, context window, modality, licence and the capability chips below are complete in both languages.

Capability legend
  • APIOfficial API
  • WebWeb search
  • ReasoningThinking mode
  • FunctionsFunction calling
  • JSONStructured output
  • OpenAI APIOpenAI-compatible API
  • CachingPrompt caching

Solid = confirmed supported; dashed = confirmed not supported; absent = not found during research.

Filters

Showing 191 / 191 models

Status
Type
Source
Ability
Country / region
Vendor

Domestic

1 countries and regions · 58 models

China

58 31 organizations

DeepSeek V4

Language

DeepSeek

Released

VersionV4-Pro / V4-Flash

Parameters
1.6T MoE (49B active)
Context window
1M
Modality
Text
License
MIT
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

MIT-licensed; reasoning and agentic ability on par with top closed models

Open weights 2026-04-24 $0.435 / $0.87

Qwen3.8-Max

Language

Alibaba

Released

Version3.8-Max

Parameters
2.4T MoE (~95B active)
Context window
1M
Modality
Text · Image
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The first open-weight Max-tier model

Open weights 2026-08-03 $2 / $6

GLM-5.3

Language

Z.ai

Released

Version5.3

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
MIT
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

State-of-the-art open coding, notably strong on cybersecurity

Open weights Date not found Not disclosed

Kimi K3

Language

Moonshot AI

Released

VersionK3

Parameters
2.8T MoE (104B active)
Context window
1M
Modality
Text · Image · Video
License
Modified MIT
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The world's first open 3T-scale model; tops the front-end coding leaderboard

Open weights 2026-07-16 $3 / $15

ByteDance

Released

Version2.1-Pro

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image
License
—
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Upgrades across three fronts: coding, agents and multimodal understanding

Closed source 2026-06-23 Not disclosed

Hunyuan Hy4

Language

Tencent

Released Tencent released the open weights on Nov 4, 2026, ending the preview phase

VersionHy4

Parameters
770B MoE (49B active)
Context window
1M
Modality
Text
License
Tencent Hunyuan License
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Open-sourced Nov 4, 2026 under Apache 2.0, with native vision and computer use

Open weights 2026-11-04 Not disclosed

ERNIE 5.1

Language

Baidu

Released

Version5.1

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image · Audio · Video
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Elastically compressed from 5.0, at just 6% of peers' pretraining cost

Closed source 2026-05-09 Not disclosed

Spark-X2.5

Language

iFlytek

Released

VersionX2.5

Parameters
293B MoE (30B active)
Context window
256K
Modality
Text
License
—
  • API
  • Reasoning
  • OpenAI API
  • Caching

Trained entirely on domestic accelerators; covers 200+ languages

Closed source 2026-09-07 Not disclosed

MiniMax M3

Language

MiniMax

Released

VersionM3

Parameters
Not disclosed (sparse MoE)
Context window
1M
Modality
Text · Image · Video
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

In-house MSA sparse attention cuts compute 20x at 1M context

Open weights 2026-06-01 Not disclosed

MiMo-V2.6-Pro

Language

Xiaomi

Released

VersionV2.6-Pro

Parameters
Trillion-parameter MoE
Context window
Not disclosed
Modality
Text · Image · Video · Audio
License
MIT
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

AA Intelligence Index 46, first among open-weight models

Open weights 2026-09-22 Not disclosed

Huawei

Released

Version2.0-Pro

Parameters
505B MoE (18B active)
Context window
512K
Modality
Text
License
OpenPangu License
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Trained natively on Ascend NPUs, deeply integrated with HarmonyOS agents

Open weights 2026-06-12 Not disclosed

Intern-S2

Language

Shanghai AI Lab

Released

VersionS2

Parameters
397B
Context window
Not disclosed
Modality
Text · Image
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • OpenAI API

Scientific ability on par with closed flagships; top tier among open models

Open weights 2026-09-13 Not disclosed

LongCat 2.0

Language

Meituan

Released

Version2.0

Parameters
1.6T MoE (~48B active)
Context window
1M
Modality
Text
License
MIT
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The first trillion-scale model trained end-to-end on 50,000 domestic accelerators

Open weights 2026-06-30 Not disclosed

Ant Group

Released

VersionLing-3.0-flash-VL

Parameters
124B MoE (5.5B active)
Context window
256K
Modality
Text · Image · Video
License
Apache-2.0
  • API
  • Reasoning

Bailing's first native multimodal model, closing the visual loop on GUIs

Open weights 2026-09-09 Not disclosed

Step 5 Preview

Language

StepFun

Preview

VersionStep 5 Preview

Parameters
600B MoE (27B active)
Context window
1M
Modality
Text · Image
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • OpenAI API
  • Caching

AA Index 44, at one-eighth the per-task cost of Opus 5

Open weights 2026-09-20 Not disclosed

MiniCPM5-2B

Language

ModelBest

Released

VersionMiniCPM5-2B

Parameters
2B
Context window
512K
Modality
Text
License
Apache-2.0
  • Reasoning
  • Functions

The strongest on-device model, just 1.2GB after INT4 quantization

Open weights 2026-09-08 Free / self-hosted

Skywork

Language

Kunlun

Released

Version3.2

Parameters
Not disclosed
Context window
Unknown
Modality
Text · Image
License
—
  • API
  • Reasoning
  • Functions
  • OpenAI API

A super agent that orchestrates hundreds of parallel sub-agents

Closed source 2026-07-01 Not disclosed

BlueLM-Pro

Language

vivo

Released

VersionPro

Parameters
Not disclosed
Context window
Unknown
Modality
Text
License
—
  • Reasoning

The cloud-side task-execution model in a device-and-cloud agent matrix

Closed source 2026-09-16 Not disclosed

AndesGPT

Language

OPPO

Released

VersionNot disclosed

Parameters
Tiny 7B / Turbo 70B / Titan (100B+)
Context window
128K
Modality
Text · Image
License
—

An on-device generative AI service with Tiny, Turbo and Titan tiers

Closed source 2026-07-15 Not disclosed

JoyAI-LLM-Flash

Language

JD.com

Released

VersionLLM-Flash

Parameters
48B MoE
Context window
Unknown
Modality
Text
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • OpenAI API

The open MoE base after JD's Yanxi became JoyAI

Open weights 2026-02-01 Not disclosed

Yuan3.0 Ultra

Language

Inspur

Released

Version3.0 Ultra

Parameters
Trillion-scale MoE
Context window
Unknown
Modality
Text · Image
License
Yuan Open License
  • API
  • Reasoning
  • Functions

One of only three trillion-scale open multimodal models worldwide

Open weights Date not found Not disclosed

ZiYue 4.0

Language

NetEase Youdao

Released

Version4.0

Parameters
27B
Context window
Unknown
Modality
Text · Image
License
Apache-2.0
  • Reasoning

Multimodal reasoning for education, with chain-of-thought compressed by 43%

Open weights 2026-05-01 Not disclosed

ByteDance

Released

Version5.0 Pro

Parameters
12B DiT
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

Major gains in image-text alignment and text rendering, with layered output

Closed source 2026-06-23 Not disclosed

Alibaba

Released

Version2.1

Parameters
7B DiT + 8B vision encoder
Context window
N/A
Modality
Text · Image
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

Native 2K with RGBA transparency; weights are non-commercial only

Open weights 2026-09-20 Not disclosed

Tencent

Preview

Version3.5 Preview

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

2K output with up to 5 reference images and multi-turn editing

Closed source 2026-09-22 Not disclosed

SenseTime

Released

VersionU1.5

Parameters
8B-MoT
Context window
N/A
Modality
Text · Image
License
Apache-2.0
  • API

Natively unified multimodal, up to 4K, strong bilingual text rendering

Open weights 2026-09-18 Not disclosed

Kuaishou

Released

VersionImage 3.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API

Native multimodal image model sharing its lineage with Video 3.0

Closed source 2026-02-05 Not disclosed

Kling 3.0

Video

Kuaishou

Released Superseded by Kling 4.0 (our news #31)

Version3.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

Native 4K at 60fps, six-shot storyboards and lip sync across five languages

Closed source 2026-02-05 Not disclosed

MiniMax H3

Video

MiniMax

Released

VersionH3

Parameters
33B Omni Transformer
Context window
N/A
Modality
Text · Image · Video · Audio
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

Native 2K stereo in a single pass; the strongest open video weights

Open weights 2026-07-31 Not disclosed

Seedance 2.5

Video

ByteDance

Released

Version2.5

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

Native 30-second clips, up to 30 reference images, extendable over multiple turns

Closed source 2026-06-23 Not disclosed

Wan 3.0

Video

Alibaba

Released

Version3.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

30 seconds per pass with native audio-video sync and structured input

Closed source 2026-08-06 Not disclosed

Vidu S2

Video

ShengShu

Released

VersionS2

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API
  • Functions
  • JSON

Real-time avatars and live video editing, continuous interaction at 720p

Closed source 2026-09-15 Not disclosed

Tencent

Released

Version1.5

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video
License
Tencent Hunyuan License
  • API
  • Functions
  • JSON

A lightweight open video generation model that runs on consumer GPUs

Open weights Date not found Free / self-hosted

Baidu

Released

Version2.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API
  • Functions
  • JSON

The first integrated Chinese audio-video model; generates multi-speaker audio in one pass

Closed source 2026-08-21 Not disclosed

iFlytek

Preview

Version1.0-Preview

Parameters
0.65B encoder + 30B MoE
Context window
N/A
Modality
Audio · Text
License
—
  • API
  • Functions
  • JSON

Recognizes 99 languages and 202 dialects, leading on noise robustness

Closed source 2026-09-16 Not disclosed

StepAudio 3

Audio

StepFun

Released

VersionStepAudio 3

Parameters
Not disclosed
Context window
N/A
Modality
Audio · Text
License
—
  • API
  • Functions
  • JSON
  • OpenAI API

Covers four tracks: ASR, real-time interaction, speech generation and music

Closed source Date not found Not disclosed

MiniMax

Released

VersionSpeech-2.8

Parameters
Not disclosed
Context window
N/A
Modality
Audio · Text
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

HD and Turbo tiers with natural emotional expression

Closed source Date not found Not disclosed

NetEase Youdao

Released

VersionConfucius4-TTS

Parameters
Not disclosed
Context window
N/A
Modality
Audio · Text
License
Apache-2.0
  • Functions
  • JSON

Accent-free cloning in 14 languages, 3-second voice copy, Apache-licensed

Open weights 2026-05-21 Free / self-hosted

Ant Group

Released

Version2.0

Parameters
100B-A6B MoE
Context window
N/A
Modality
Audio · Text
License
Apache-2.0
  • API
  • Functions
  • JSON

Unified audio generation for all scenarios — speech, sound effects and music in one track

Open weights 2026-02-11 Free / self-hosted

Qwen3-Embedding 8B

Embedding

Alibaba

Released

Version8B

Parameters
0.6B / 4B / 8B
Context window
32K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

The ceiling for open retrieval quality, with strong multilingual performance

Open weights Date not found Free / self-hosted

BGE-M3

Embedding

BAAI

Released

VersionM3

Parameters
568M
Context window
8K
Modality
Text
License
MIT
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Dense, sparse and multi-vector in one; the default base for multilingual RAG

Open weights Date not found Free / self-hosted

Tencent

Released

VersionHy World 2.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · 3D
License
Tencent Hunyuan License
  • Web
  • Reasoning
  • Functions
  • JSON

Generates and reconstructs 3D worlds from text, images and video, exporting assets in multiple formats

Open weights 2026-04-16 Free / self-hosted

Kunlun

Released

VersionMatrix-3D

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · 3D
License
Apache-2.0
  • Web
  • Reasoning
  • Functions
  • JSON

Panoramic representations that generate walkable 3D worlds

Open weights Date not found Free / self-hosted

DeepSeek

Released Current API snapshot is V4.1-Flash-1017; legacy alias retired

Version4.1

Parameters
552B MoE (8B/16B active)
Context window
1M
Modality
Text · Image
License
MIT
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

First to use an asymmetric Causal-Encoder-Decoder architecture, cutting the KV cache to a quarter of the previous generation; the weights are open, and peak-hour usage is billed at 2x

Open weights 2026-09-10 $0.15 / $0.6

Moonshot AI

Preview Official preview build, not yet GA

VersionK2.8 Preview

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
  • API
  • Reasoning
  • Functions
  • OpenAI API
  • Caching

A mid-tier model sitting between K2.7 Code and the flagship K3, with overall performance close to K3; the 1M context window is opened to all members (pricing from third-party gateways)

Closed source 2026-09-11 $0.8 / $3.35

ByteDance

Released

Version5.0 Flash

Parameters
Not disclosed
Context window
—
Modality
Text · Image
License
—
  • API
  • OpenAI API

The fast tier of the Seedream 5.0 family: splits an image into up to 16 editable layers, supports coordinate and sketch-mark editing, and blends up to 10 reference images

Closed source 2026-10-01 Not disclosed

Yi-34B

Language

01.AI

Released

Version34B

Parameters
34B
Context window
32K
Modality
Text
License
Apache-2.0
  • API
  • Reasoning
  • OpenAI API

A bilingual 34B dense model that topped open-source leaderboards at release.

Open weights 2023-11-06 Free / self-hosted

Baichuan Intelligence

Released

Version2 / 13B

Parameters
13B
Context window
4K
Modality
Text
License
Baichuan 2 Community License
  • Reasoning
  • Functions

A bilingual open dialogue model with notably strong Chinese, including medical and legal domains.

Open weights 2023-09-06 Free / self-hosted

GLM-4-9B

Language

Zhipu AI

Released

Version4 / 9B

Parameters
9B
Context window
128K
Modality
Text
License
GLM-4 License
  • API
  • Web
  • Reasoning
  • Functions
  • OpenAI API

The open-weights branch of ChatGLM: strong Chinese, 128K context and tool calling.

Open weights 2024-06-05 Free / self-hosted

Qwen3-235B-A22B

Language

Alibaba

Released

Version3 / 235B-A22B

Parameters
235B MoE (22B active)
Context window
128K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Alibaba’s flagship open-weights MoE with a native hybrid thinking mode and 119 languages.

Open weights 2025-04-29 Free / self-hosted

Alibaba

Released

Version2.5-VL / 32B

Parameters
32B
Context window
128K
Modality
Text · Image · Video
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • OpenAI API

An open vision-language model with native dynamic resolution; strong at documents and video.

Open weights Date not found Free / self-hosted

CogVLM2

Vision

Zhipu AI

Released

VersionCogVLM2 / 19B

Parameters
19B
Context window
8K
Modality
Text · Image
License
CogVLM2 License
  • API
  • Reasoning

A high-scoring open vision model, strong at image-text and OCR understanding.

Open weights Date not found Free / self-hosted

SenseTime

Released

VersionV6.5-Pro

Parameters
Not disclosed
Context window
—
Modality
Text · Image · Video
License
—
  • API
  • OpenAI API

SenseTime’s natively multimodal flagship, unifying text, image and video.

Closed source Date not found Not disclosed

APUS-OpenJev

Decision

APUS

Released

Versionv1(9B / 4B)

Parameters
9B / 4B
Context window
—
Modality
Text
License
MIT
  • API
  • Web
  • Reasoning

APUS’s independent open reproduction of Jev, packaged as an offline-runnable Agent Skill.

Open weights 2026-09-19 Free / self-hosted

IRIS

Language

Xiaohongshu

Released

Version35B / 397B

Parameters
35B / 397B
Context window
—
Modality
Text
License
Apache-2.0
  • Web

Xiaohongshu AllSpark’s open retrieval-augmented agent model, in 35B and 397B sizes.

Open weights 2026-09-14 Free / self-hosted

MiniMax

Released Live on the official site (marked NEW); no release date published

VersionM3.1 Flash

Parameters
—
Context window
—
Modality
Text
License
—

Flagged NEW on the official model nav; no explicit release date

Open weights Date not found Not disclosed

MiniMax

Released Open weights published; date not announced

Version3.0

Parameters
—
Context window
—
Modality
Audio
License
—

Marked open-weights and production-ready on the official site

Open weights Date not found Not disclosed

Seedream 5.5

Image

ByteDance

Released Released Nov 9, 2026 with API availability

Version5.5

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API

Released by ByteDance Seed on Nov 9, 2026, focused on pixel-stable multi-round editing

Closed source 2026-11-09 Not disclosed

International

18 countries and regions · 133 models

United States

88 36 organizations

GPT-6 Astra

Language

OpenAI

Released

Version6 Astra

Parameters
Not disclosed
Context window
1.05M
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Best at computer use and long-horizon agents; Sol and Luna siblings available

Closed source 2026-09-04 $10 / $50

OpenAI

Released

Version2.5(Flare / Sunburst)

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Functions
  • JSON
  • OpenAI API
  • Caching

Strong text rendering and multi-turn editing, 50% faster generation

Closed source 2026-09-08 $5 / $30

OpenAI

Released

Version2.1

Parameters
Not disclosed
Context window
128K
Modality
Text · Audio
License
—
Knowledge cutoff
2024-09-30
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The first real-time voice dialogue model with GPT-5-level reasoning

Closed source 2026-08-01 $32 / $64

OpenAI

Released

Version3 large

Parameters
Not disclosed
Context window
8191
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

3072 dimensions; the default choice for English retrieval

Closed source 2024-01-25 $0.13 / —

Claude Opus 5.5

Language

Anthropic

Released

VersionOpus 5.5

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image
License
—
Knowledge cutoff
2026-06-01
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The most capable Opus, priced 40% below Opus 5

Closed source 2026-09-22 Not disclosed

Claude Fable 5.1

Language

Anthropic

Released

VersionFable 5.1

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
Knowledge cutoff
2026-06-01
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Best for coding and research; Mythos 5.1 is its restricted sibling

Closed source 2026-09-01 $10 / $50

Claude Sonnet 5

Language

Anthropic

Released

VersionSonnet 5

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The most agentic Sonnet, at a strong price-performance ratio

Closed source 2026-06-30 $2 / $10

Gemini 3.1 Pro

Language

Google DeepMind

Released

Version3.1 Pro

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image · Audio · Video
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Flagship Pro tier leading 12 of 18 benchmarks

Closed source 2026-02-01 $2 / $12

Gemini 3.8 Flash

Language

Google DeepMind

Released

Version3.8 Flash

Parameters
Not disclosed
Context window
1M
Modality
Text · Image · Audio · Video
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The current workhorse, with adjustable effort to balance cost and quality

Closed source 2026-09-02 $0.75 / $3.75

Google DeepMind

Released

Version3.8 Live / Live Extended Thinking

Parameters
Not disclosed
Context window
Not disclosed
Modality
Audio · Text · Image
License
—
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

First on the speech-quality leaderboard; speaks while it reasons

Closed source 2026-09-15 Not disclosed

Google DeepMind

Released

VersionPro(Gemini 3 Pro Image)

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Best-in-class 4K and multilingual text rendering

Closed source 2025-11-20 $2 / $12

Veo 3.1

Video

Google DeepMind

Released

Version3.1

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Native synchronized audio with cinematic camera control

Closed source Date not found Not disclosed

Lyria 3.5

Audio

Google DeepMind

Released

Version3.5

Parameters
Not disclosed
Context window
N/A
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Generates complete 3-minute songs with vocals, at about $0.08 each

Closed source 2026-09-04 Not disclosed

Gemma 4

Language

Google DeepMind

Released

Version4

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image
License
Gemma Terms of Use
  • API
  • Web
  • Reasoning
  • Functions
  • JSON

Flagship open weights, emphasizing intelligence density per parameter

Open weights Date not found Free / self-hosted

Gemini Embedding 2

Embedding

Google DeepMind

Released

Version2

Parameters
Not disclosed
Context window
8192
Modality
Text · Image · Audio · Video
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Unified multimodal vector space, adjustable from 128 to 3072 dimensions

Closed source 2026-04-01 $0.2 / —

Grok 4.7

Language

xAI

Released

Version4.7

Parameters
Not disclosed
Context window
500K
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Built for coding and agents, with low hallucination rates and aggressive pricing

Closed source 2026-09-21 $2 / $6

xAI

Released

Version2.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Inpainting with multiple reference images, about $0.04 per image

Closed source 2026-08-07 Not disclosed

xAI

Released

Version1.5

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

6-second 720p with native audio, rendered in about 25 seconds

Closed source 2026-06-16 Not disclosed

Muse Spark 1.3

Language

Meta

Released

Version1.3

Parameters
Not disclosed
Context window
1.05M
Modality
Text · Image · Video · Audio
License
—
  • API
  • Reasoning
  • Functions

Built for long agentic tasks; output priced below peer flagships

Closed source 2026-09-02 $1.25 / $4.25

Muse Glimmer 30B

Language

Meta

Released

Version30B

Parameters
29.6B
Context window
131K
Modality
Text · Image
License
Llama Community License
Knowledge cutoff
2026-01-04
  • Functions
  • OpenAI API

Apache 2.0; a local agent that runs on 24GB of VRAM

Open weights 2026-08-10 Free / self-hosted

NVIDIA

Released

Version3 Ultra 550B-A55B

Parameters
550B (55B active)
Context window
Not disclosed
Modality
Text
License
NVIDIA Open Model License
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

The largest US open-weight reasoning model

Open weights 2026-06-04 Free / self-hosted

NVIDIA

Released

Version3.5 Lightning

Parameters
30B (3B active)
Context window
1M
Modality
Text
License
NVIDIA Open Model License
  • API
  • Functions
  • OpenAI API

Runs on a single laptop GPU; fastest agent throughput

Open weights 2026-08-11 Free / self-hosted

NVIDIA

Released

Version2.0

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
NVIDIA Open Model License
  • Reasoning
  • Functions
  • JSON

Generates a walkable 3D world from one image, exportable as mesh or 3DGS

Open weights 2026-04-15 Free / self-hosted

MAI-Thinking 1

Language

Microsoft

Released

VersionThinking-1

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text
License
—
  • API
  • Reasoning
  • Functions
  • OpenAI API

Enterprise mid-size reasoning, trained in-house rather than distilled

Closed source 2026-08-13 Not disclosed

Microsoft

Released

Version2.5

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Text-to-image generation and editing, plus a faster Flash tier

Closed source Date not found Not disclosed

MAI-Voice 2

Audio

Microsoft

Released

Version2

Parameters
Not disclosed
Context window
N/A
Modality
Text · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Low-latency, highly expressive speech synthesis

Closed source Date not found Not disclosed

Amazon

Released

Version2 Lite

Parameters
Not disclosed
Context window
1M
Modality
Text · Image · Video
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • OpenAI API

Bedrock's value workhorse, with a built-in code interpreter

Closed source 2026-04-01 Not disclosed

Amazon

Released

Version2 Sonic

Parameters
Not disclosed
Context window
N/A
Modality
Text · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Low-latency real-time speech-to-speech

Closed source Date not found Not disclosed

Apple

Released

VersionAFM 3(Core / Core Advanced)

Parameters
3B / 20B (1–4B active)
Context window
Not disclosed
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

An on-device multimodal model with zero cost and zero upload locally

Closed source 2026-06-08 Free / self-hosted

Allen Institute for AI

Released

Version3.1 Think 32B

Parameters
32B
Context window
64K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Full training pipeline open-sourced; one of the strongest open thinking models

Open weights Date not found Free / self-hosted

IBM Granite 4.1

Language

IBM

Released

Version4.1

Parameters
Not disclosed
Context window
32K
Modality
Text · Image · Audio
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • OpenAI API

Apache 2.0 enterprise-grade, with speech and embedding variants

Open weights 2026-04-29 Free / self-hosted

Perplexity

Released

VersionSonar 2

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text
License
—
  • API
  • Web
  • Functions
  • OpenAI API

Built-in web retrieval and citations, purpose-built for search

Closed source Date not found Not disclosed

LFM2.5-2.6B

Language

Liquid AI

Released

VersionLFM2.5-2.6B

Parameters
2.69B
Context window
131K
Modality
Text
License
LFM Open Weights License
  • Web
  • Functions
  • OpenAI API

An on-device agent matching models four times its size

Open weights 2026-08-04 Free / self-hosted

Cartesia

Released

VersionSonic 3.6

Parameters
Not disclosed
Context window
N/A
Modality
Text · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • Caching

First audio under 90ms; tops the real-time speech quality leaderboard

Closed source 2026-08-01 Not disclosed

ElevenLabs

Released

Versionv3

Parameters
Not disclosed
Context window
N/A
Modality
Text · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Best emotion control; the first choice for long-form audiobooks and dubbing

Closed source 2025-06-01 Not disclosed

Runway

Released

VersionGen-4.5

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • Caching

Prompt-driven multi-shot camera movement control

Closed source Date not found Not disclosed

Luma Ray 3.2

Video

Luma AI

Released

VersionRay 3.2

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • Caching

Up to 16 keyframes with HDR and EXR output

Closed source Date not found Not disclosed

Midjourney

Released

VersionV8.2

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Best-in-class aesthetics and photorealism; no official API

Closed source 2026-07-24 Not disclosed

Suno v5.5

Audio

Suno

Released

Versionv5.5

Parameters
Not disclosed
Context window
N/A
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • Caching

One-click full songs; the most natural vocals and arrangement

Closed source 2026-03-26 Not disclosed

GPT-6.1 Sol

Language

OpenAI

Released

Version6.1

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
Knowledge cutoff
2026-04-30
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

The mid-tier workhorse of the GPT-6 family announced at DevDay 2026: agentic coding and computer use close to GPT-6 Astra at a fifth of the price

Closed source 2026-09-29 $2 / $10

GPT-6 Luna

Language

OpenAI

Released

Version6

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
  • API
  • OpenAI API

The ultra-low-cost tier of the GPT-6 family at $0.10/$0.50 per million tokens, aimed at large-scale lightweight tasks

Closed source 2026-09-22 $0.1 / $0.5

Gemini 4 Argon

Language

Google DeepMind

Released

Version4

Parameters
Not disclosed
Context window
1M
Modality
Text · Image · Video
License
—
  • API
  • Reasoning
  • Caching

Google's new flagship and the first Gemini 4 model: long-horizon reasoning plus cyber-defence, with the output ceiling raised to roughly 1M tokens; released in stages under the Fairwind programme

Closed source 2026-09-30 $2 / $10

Anthropic

Released

Version5.5

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
Knowledge cutoff
2026-06-01
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

More than 30% faster than Sonnet 5 and up to 30% cheaper per task, with Terminal-Bench 4.0 jumping from 10.3% to 70.6%; already used in the free Claude tier

Closed source 2026-09-28 $2 / $10

IBM Granite 4.2

Language

IBM

Released

Version4.2

Parameters
3B / 8B / 30B (dense)
Context window
128K (30B extends to 512K)
Modality
Text
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • OpenAI API

An open dense reasoning family built for enterprise agents, with a native thinking mode and reasoning-enhanced tool calling; the 8B and 30B sizes were RL-trained in real environments

Open weights 2026-08-25 Not disclosed

Llama 4 Maverick

Language

Meta

Released

Version4 Maverick

Parameters
400B MoE (17B active)
Context window
1M
Modality
Text · Image
License
Llama 4 Community License
Knowledge cutoff
2024-08-01
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Meta’s first natively multimodal open-weights MoE — 128 experts, 17B active parameters.

Open weights 2025-04-05 Free / self-hosted

Llama 4 Scout

Language

Meta

Released

Version4 Scout

Parameters
109B MoE (17B active)
Context window
10M
Modality
Text · Image
License
Llama 4 Community License
Knowledge cutoff
2024-08-01
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

A 10M-token context window; runs on a single H100 once quantised.

Open weights 2025-04-05 Free / self-hosted

Phi-4

Language

Microsoft

Released

Version4

Parameters
14B
Context window
16K
Modality
Text
License
MIT
Knowledge cutoff
2024-06-01
  • API
  • Reasoning
  • OpenAI API

A quality benchmark for small models: trained on synthetic textbook data, strong at maths and reasoning.

Open weights 2024-12-12 Free / self-hosted

OpenAI

Released

Versionlarge-v3

Parameters
1.55B
Context window
30 s audio window
Modality
Text · Audio
License
MIT
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

The open speech-recognition benchmark: transcription and translation across 99 languages.

Open weights 2023-11-06 Free / self-hosted

LLaVA-1.6

Vision

LLaVA Team

Released

Version1.6 / 34B

Parameters
34B
Context window
4K
Modality
Text · Image
License
Apache-2.0
  • Web
  • Reasoning
  • Functions
  • OpenAI API

The defining open visual-instruction model — the classic CLIP + LLM baseline.

Open weights 2024-01-30 Free / self-hosted

Nomic AI

Released

Versionv1.5

Parameters
137M
Context window
8192
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

A fully reproducible text embedding model with an 8192-token context.

Open weights Date not found Free / self-hosted

Microsoft

Released

Versionmistral-7b-instruct

Parameters
7B
Context window
32K
Modality
Text
License
MIT
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

An open embedding model built on Mistral-7B, strong at long-context retrieval.

Open weights Date not found Free / self-hosted

DBRX Instruct

Language

Databricks

Released

VersionInstruct / 132B

Parameters
132B MoE (36B active)
Context window
32K
Modality
Text
License
Databricks Open Model License
  • Web
  • Reasoning
  • Functions
  • OpenAI API

Databricks’ open MoE model, strong at code and SQL generation.

Open weights 2024-03-27 Free / self-hosted

Snowflake

Released

VersionInstruct / 480B

Parameters
480B MoE (17B active)
Context window
4K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning

An enterprise-grade open MoE under Apache-2.0, notable for code and SQL.

Open weights 2024-04-24 Free / self-hosted

gpt-oss-120b

Language

OpenAI

Released

Version120b

Parameters
117B MoE (5.1B active)
Context window
128K
Modality
Text
License
Apache-2.0
Knowledge cutoff
2024-06-01
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

OpenAI’s first open-weights model; runs on a single 80GB GPU.

Open weights 2025-08-05 Free / self-hosted

Claude Haiku 4.5

Language

Anthropic

Released Superseded by Haiku 5.5 on 2026-10-07; still callable

VersionHaiku 4.5

Parameters
Not disclosed
Context window
200K
Modality
Text · Image
License
—
Knowledge cutoff
2025-02-28
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Anthropic’s fastest and cheapest tier, aimed at real-time and high-concurrency agent workloads.

Closed source 2025-10-15 $1 / $5

Palmyra X5

Language

Writer

Released

VersionX5

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
  • API
  • Reasoning
  • Functions
  • JSON

An enterprise long-context model with a 1M-token window, built around multi-agent workflows.

Closed source Date not found $0.6 / $6

Reka Core

Language

Reka AI

Released

VersionCore

Parameters
Not disclosed
Context window
128K
Modality
Text · Image · Audio · Video
License
—
  • API
  • Functions
  • JSON
  • OpenAI API

A natively multimodal frontier model: text, image, video and audio understanding behind one API.

Closed source Date not found $2 / $6

Google Cloud

Released

VersionChirp 3

Parameters
Not disclosed
Context window
—
Modality
Audio · Text
License
—
  • API
  • OpenAI API

Google Cloud’s flagship speech-recognition model, with streaming and dynamic batching.

Closed source Date not found Not disclosed

AssemblyAI

Released

VersionUniversal-3.5 Pro

Parameters
Not disclosed
Context window
—
Modality
Audio · Text
License
—
  • API
  • OpenAI API

High-accuracy transcription with speaker diarisation and domain modes such as medical.

Closed source Date not found Not disclosed

Deepgram

Released

VersionNova-3

Parameters
Not disclosed
Context window
—
Modality
Audio · Text
License
—
  • API
  • OpenAI API

Leading low-latency streaming recognition at a 5.26% word error rate, with medical and finance lines.

Closed source Date not found Not disclosed

Ideogram 3.0

Image

Ideogram

Released

Version3.0

Parameters
Not disclosed
Context window
—
Modality
Text · Image
License
—
  • API
  • OpenAI API

The benchmark for text rendered inside images; strong at layout and branded content.

Closed source Date not found Not disclosed

Recraft V3

Image

Recraft

Released

VersionV3

Parameters
Not disclosed
Context window
—
Modality
Text · Image
License
—
  • API
  • Functions
  • JSON
  • OpenAI API
  • Caching

A design-oriented image model with a rare native SVG (vector) output mode.

Closed source 2024-10-01 Not disclosed

Hedra

Released

VersionCharacter 3

Parameters
Not disclosed
Context window
—
Modality
Text · Image · Audio · Video
License
—
  • API
  • OpenAI API

One portrait plus audio yields high-fidelity talking and singing video, up to 10 minutes long.

Closed source Date not found Not disclosed

Pika 2.5

Video

Pika

Released

Version2.5

Parameters
Not disclosed
Context window
—
Modality
Text · Image · Video
License
—
  • API
  • OpenAI API

Effects-first short-video generation, with Pikaffects object transformations and more.

Closed source Date not found Not disclosed

Microsoft

Released

VersionNeural HD 2.5

Parameters
Not disclosed
Context window
—
Modality
Text · Audio
License
—
  • API
  • OpenAI API

An enterprise speech platform: synthesis, recognition, translation and custom voices.

Closed source Date not found Not disclosed

Amazon Polly

Audio

Amazon

Released

VersionStandard / Neural / Generative / Long-Form

Parameters
Not disclosed
Context window
—
Modality
Text · Audio
License
—
  • API
  • OpenAI API

AWS-managed TTS with four quality tiers: standard, neural, generative and long-form.

Closed source Date not found Not disclosed

Jev

Decision

TypeSafe AI

Released

Versionjev-1.13.0

Parameters
—
Context window
64k tokens per request
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions

TypeSafe AI's System One decision model: no text output, just typed answers with calibrated probabilities; the SDK and Agent Skill are MIT-licensed.

Closed source 2026-09-15 $0.042 / $0

Cloudflare Clef

Decision

Cloudflare

Released

Version27B

Parameters
27B (frozen Qwen3.8-27B + routing head + LoRA)
Context window
65536
Modality
Text · Image · Video
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions

Cloudflare’s own decision model, API-compatible with Jev and able to read images too.

Open weights 2026-10-01 $0.24 / $0

Cloudflare

Released

Version9B

Parameters
9B (frozen Qwen3.5-9B + routing head + LoRA)
Context window
65536
Modality
Text · Image · Video
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions

The 9B lightweight Clef: a decision in about 40ms, the cheapest open option.

Open weights 2026-10-01 $0.09 / $0

Laya

Decision

Convai Innovations

Released

Version421M

Parameters
~421M
Context window
512
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions

An open decision model at roughly 400M parameters that self-hosts on CPU and at the edge.

Open weights 2026-09-22 Free / self-hosted

pplx-decider

Decision

Perplexity

Released

Versionv1-27b

Parameters
27B
Context window
—
Modality
Text · Image
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions

Perplexity’s multimodal decision model: probabilities from text or images, with open weights.

Open weights Date not found $0.04 / $0

Mercury Decide

Decision

Inception

Released

VersionMercury Decide

Parameters
—
Context window
33k
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions

Inception’s decision endpoint using Jev’s question types, free to try.

Closed source 2026-09-30 Not disclosed

Claude Haiku 5.5

Language

Anthropic

Released Announced 2026-10-07; GA in the Claude API

Version5.5

Parameters
—
Context window
1M
Modality
Text · Image
License
—
Knowledge cutoff
2026-06-01

One of the four current Claude models; retirement commitment no earlier than 2027-10-07

Closed source 2026-10-07 $0.1 / $0.5

Claude Mythos

Language

Anthropic

Preview Not publicly available; requires the LSVP verification program

VersionMythos(5.1 代)

Parameters
—
Context window
1M
Modality
Text · Image
License
—

Listed as accessible under the Life Sciences Verification Program; no standalone launch

Closed source 2026-09-15 Not disclosed

Claude Fable 5.5

Language

Anthropic

Unconfirmed Unannounced; leaked via cloud service config

Version5.5

Parameters
—
Context window
—
Modality
Text · Image
License
—

Unannounced; found in regional cloud service records. Covered in our news #28

Closed source Date not found Not disclosed

GPT-5.6 Cyber

Language

OpenAI

Preview Gated to authorized security researchers

Version5.6 Cyber

Parameters
—
Context window
—
Modality
Text
License
—

Listed under OpenAI Daybreak; requires authorization

Closed source Date not found Not disclosed

GPT-Rosalind

Language

OpenAI

Preview Requires institutional approval

VersionRosalind

Parameters
—
Context window
—
Modality
Text
License
—

OpenAI docs list it as available to approved institutions

Closed source Date not found Not disclosed

OpenAI

Unconfirmed Unannounced; surfaced via blind arena testing

VersionRed / Green / Alpha

Parameters
—
Context window
—
Modality
Image
License
—

Unannounced; spotted in blind arena tests. Covered in our news #32

Closed source Date not found Not disclosed

GPT-6.1 Terra

Language

OpenAI

Released Generally available Oct 26, 2026 in ChatGPT Work, Codex and API (gpt-6.1-terra)

Version6.1

Parameters
—
Context window
1.05M
Modality
Text · Image
License
—
Knowledge cutoff
2026-06-01
  • Web
  • Reasoning
  • Functions
  • Caching

Supports computer use, web search and functions; reasoning effort none–max; knowledge cutoff Jun 2026

Closed source 2026-10-26 $1 / $5

Genie 4

3D

Google DeepMind

Preview Research preview with a limited group of trusted testers; AI Ultra subscribers and early access partners are next

Version4

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • Reasoning

Built on Gemini Omni plus Gemini reasoning models; audio and dialogue generated on the fly

Closed source 2026-10-27 Free / self-hosted

Gemini 4 Carbon

Language

Google DeepMind

Released Generally available since Nov 10, 2026; the Oct 28 launch was limited to cyber defenders and Fairwind partners

Version4 Carbon

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • Caching

Ranks #1 on Artificial Analysis Intelligence Index v5.0 (59); SEC-bench Pro 88.2%, CWE-bench 71.0%

Closed source 2026-10-28 $2 / $10

OpenAI

Released Live since Nov 4, 2026 across ChatGPT, Codex and the API, on every subscription tier

Version2.6(Flare / Sunburst)

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image
License
—
  • API
  • OpenAI API
  • Caching

Flare trades for speed, Sunburst for detail and cross-edit consistency; adds complexity and quality controls

Closed source 2026-11-04 $5 / $30

GPT-6 Cyber

Language

OpenAI

Preview No self-serve access; available only through the Daybreak trusted-access programme (Daybreak Red)

Version6 Cyber

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text
License
—
  • API
  • Reasoning
  • Functions

ExploitGym 48.5% vs 32.1% for GPT-5.6 Cyber; classified Critical under the Preparedness Framework

Closed source 2026-11-05 Not disclosed

Google DeepMind

Released Generally available Nov 5, 2026 on the Gemini API, Google Cloud, Google Vids and Gemini for business

Version1.2 Flash

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API

First generally available Gemini Omni release; supports character, product and style references with a SynthID watermark

Closed source 2026-11-05 Not disclosed

Muse Ember

Language

Meta

Released Announced by Meta AI on Nov 9, 2026

VersionEmber

Parameters
Not disclosed
Context window
1M
Modality
Text · Image
License
—
  • Reasoning
  • Functions

1M context, aimed at high-frequency repetitive agent subtasks while larger models do the judgement

Closed source 2026-11-09 Not disclosed

Claude Fable 5.2

Language

Anthropic

Released Available to all users since Nov 9, 2026, with cyber and biology safeguards

VersionFable 5.2

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • Caching

FrontierCode 2.0 Xhigh 49.1%, highest of any released model; HLE with tools 74.2%; can serve as Advisor or Orchestrator

Closed source 2026-11-09 $8 / $40

Anthropic

Preview Requires approval through the Cyber or Life Sciences Verification Program; no self-serve access

VersionMythos 5.2

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text · Image
License
—
  • API
  • Reasoning
  • Functions

CyScenarioBench 76.3%, CVE Rediscovery 89.6%; verified defensive and research use only

Closed source 2026-11-09 $8 / $40

Google DeepMind

Released Rolled out globally Nov 12, 2026 across the Gemini API, AI Studio, the Gemini app, Search and Antigravity

Version4 Hydrogen

Parameters
Not disclosed
Context window
1M
Modality
Text · Image · Video · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • Caching

AA Intelligence Index 51 (default high) / 54 (xhigh) vs Argon 50; Terminal-Bench 4.1 57% vs 18.2% for 3.8 Flash

Closed source 2026-11-12 $1.5 / $7.5

France

9 1 organizations

Mistral Large 3

Language

Mistral AI

Released Superseded by Mistral Large 4 on 2026-10-06

VersionLarge 3(25.12)

Parameters
675B MoE (41B active)
Context window
256K
Modality
Text · Image
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Europe's strongest open-weight general multimodal flagship

Open weights 2025-12-02 Not disclosed

Mistral AI

Released

VersionMedium 3.5(26.04)

Parameters
128B
Context window
256K
Modality
Text · Image
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Mistral's current API flagship, unifying reasoning and coding

Open weights 2026-05-22 Not disclosed

Mistral Small 4

Language

Mistral AI

Released

VersionSmall 4(26.03)

Parameters
119B MoE
Context window
256K
Modality
Text · Image
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

An efficient open model merging reasoning, vision and coding

Open weights 2026-03-16 Not disclosed

Mistral AI

Released

VersionVoxtral TTS(26.03)

Parameters
4B
Context window
N/A
Modality
Text · Audio
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Open zero-shot TTS that clones a voice in 5 seconds across nine languages

Open weights 2026-03-23 Not disclosed

Mixtral 8x7B

Language

Mistral AI

Released

Version8x7B

Parameters
46.7B MoE (12.9B active)
Context window
32K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

The first open-weights MoE model; fast inference and a mature deployment ecosystem.

Open weights 2023-12-11 Free / self-hosted

Mistral Large 4

Language

Mistral AI

Released Released 2026-10-06, supersedes Mistral Large 3

Version4

Parameters
—
Context window
—
Modality
Text
License
—

Announced as "Introducing Mistral Large 4" on 2026-10-06

Closed source 2026-10-06 Not disclosed

Mistral AI

Released Released 2026-10-20

Version3

Parameters
—
Context window
—
Modality
Audio · Text
License
—

Released 2026-10-20; third generation of the Voxtral speech line

Closed source 2026-10-20 Not disclosed

Mistral AI

Released Released 2026-10-15 as open weights

Version209M

Parameters
209M
Context window
—
Modality
Text
License
—

Released 2026-10-15; probabilistic forecasting model

Open weights 2026-10-15 Not disclosed

Mistral AI

Released Released 2026-08-04

Version1.0

Parameters
—
Context window
—
Modality
Text
License
—

Released 2026-08-04 under Solutions

Open weights 2026-08-04 Not disclosed

Canada

5 1 organizations

Cohere

Released

VersionA+

Parameters
218B MoE (25B active)
Context window
128K
Modality
Text · Image
License
CC-BY-NC-4.0
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Enterprise MoE flagship that self-hosts on a single B200

Open weights 2026-05-01 Free / self-hosted

Cohere Embed 4

Embedding

Cohere

Released

Versionembed-v4.0

Parameters
Not disclosed
Context window
128K
Modality
Text · Image
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Multimodal retrieval embeddings; the go-to for RAG

Closed source 2025-04-01 $0.12 / —

Cohere Rerank 4

Embedding

Cohere

Released

Versionrerank-v4.0(pro / fast)

Parameters
Not disclosed
Context window
32K
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

RAG reranking, billed per search

Closed source Date not found Not disclosed

Cohere Embed 5

Embedding

Cohere

Released

Version5(Pro / Fast)

Parameters
Not disclosed
Context window
128K
Modality
Text · Image
License
—
  • API

A multimodal, multilingual embedding family (Pro $0.12 / Fast $0.08 per million tokens); Pro and Fast share one vector space so you can switch without re-indexing, and it scores 85.8 on ViDoRe V3

Closed source 2026-09-30 $0.12 / —

Aya Expanse 32B

Language

Cohere

Released

VersionExpanse 32B

Parameters
32B
Context window
128K
Modality
Text
License
CC-BY-NC-4.0
  • API
  • Reasoning
  • Functions

Cohere’s open multilingual model covering 23 languages.

Open weights 2024-10-24 Free / self-hosted

Germany

5 4 organizations

Aleph Alpha

Released

Version7B control

Parameters
7B
Context window
8K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

A European sovereign-compliance model focused on auditability and control

Open weights 2024-08-01 Not disclosed

Black Forest Labs

Released

Version3

Parameters
Not disclosed
Context window
N/A
Modality
Text · Image · Video · Audio
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Unified image, video and audio multimodal, with synchronized audio in video

Closed source 2026-07-23 Not disclosed

DeepL Translator

Language

DeepL

Released

VersionNext-generation LLM translation engine

Parameters
Not disclosed
Context window
Unknown
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

A translation-specific model, first in quality for European language pairs

Closed source 2024-07-01 Not disclosed

Kolibri

Language

Aleph Alpha

Released

Version1

Parameters
78B MoE (3B active)
Context window
1M
Modality
Text
License
Apache-2.0

An open bilingual English-German MoE model focused on long context and abstention for regulated, mission-critical settings

Open weights 2026-10-03 Not disclosed

jina-embeddings-v3

Embedding

Jina AI

Released

Versionv3

Parameters
570M
Context window
8192
Modality
Text
License
CC-BY-NC-4.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

A multilingual embedding model using Task LoRA for retrieval, classification and clustering.

Open weights 2024-09-18 Free / self-hosted

Japan

4 4 organizations

NTT tsuzumi 2

Language

NTT

Released

Version2

Parameters
~30B
Context window
Not disclosed
Modality
Text · Image
License
—
  • API
  • Reasoning

A lightweight Japanese-native LLM that runs on a single GPU

Closed source 2025-10-01 Not disclosed

SB Intuitions

Released

Version3 mini

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text
License
—
  • API

Japanese-native, selected for the Digital Agency's government pilot

Closed source 2026-06-30 Not disclosed

Sakana AI

Released

VersionUltra v2

Parameters
Not disclosed (multi-model orchestration)
Context window
~1M
Modality
Text · Image
License
—
  • Reasoning

Dynamic multi-model orchestration; not yet available in the EU

Closed source 2026-09-11 Not disclosed

Rakuten AI 3.0

Language

Rakuten

Released

Version3.0

Parameters
671B MoE (37B active)
Context window
Not disclosed
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning

Japan's largest open model, built on a DeepSeek-V3 base

Open weights 2026-03-17 Free / self-hosted

South Korea

4 4 organizations

LG EXAONE 4.5

Language

LG AI Research

Released

Version4.5

Parameters
33B (incl. 1.2B vision encoder)
Context window
256K
Modality
Text · Image
License
EXAONE License
  • API
  • Web
  • Reasoning

LG's first open vision-language model, strong on document understanding

Open weights 2026-04-09 Free / self-hosted

Upstage

Released

VersionPro 4

Parameters
Not disclosed
Context window
524K
Modality
Text
License
—
  • API
  • Reasoning
  • OpenAI API
  • Caching

Korea's first flagship model to appear on OpenRouter

Closed source 2026-08-14 $0.3 / $1.2

SK Telecom

Released

VersionK2

Parameters
688B MoE (33B active)
Context window
256K
Modality
Text
License
Apache-2.0
  • API

Korea's largest open MoE flagship

Open weights 2026-07-29 Free / self-hosted

United Arab Emirates

3 2 organizations

TII

Released

VersionH1 Arabic(34B)

Parameters
3B / 7B / 34B
Context window
256K
Modality
Text
License
TII Falcon License
  • API
  • Web
  • Reasoning

First among open Arabic models; tops the OALL leaderboard

Open weights 2026-01-05 Free / self-hosted

TII

Released

VersionPerception 0.6B

Parameters
0.6B
Context window
N/A
Modality
Image
License
TII Falcon License
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Early-fusion visual segmentation and OCR, more efficient than SAM3

Open weights 2026-03-31 Free / self-hosted

K2-65B

Language

MBZUAI

Released

Version65B

Parameters
65B
Context window
4K
Modality
Text
License
Apache-2.0
  • API
  • Web
  • Reasoning

A fully reproducible open model: training details and intermediate checkpoints are all public.

Open weights 2024-06-03 Free / self-hosted

Russia

2 2 organizations

Yandex

Released

Version5.1 Pro

Parameters
Not disclosed
Context window
32K
Modality
Text
License
—
  • API
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Best in Russian, served through Yandex Cloud AI Studio

Closed source 2025-08-01 Not disclosed

Sber

Released

Version3.5 Reasoning

Parameters
432B MoE (28B active)
Context window
262K
Modality
Text
License
Apache-2.0
  • API
  • Reasoning
  • Functions
  • JSON

Russia's first open linear-attention MoE, MIT-licensed

Open weights 2026-09-10 Not disclosed

India

2 2 organizations

Sarvam 105B

Language

Sarvam AI

Released

Version105B

Parameters
105B MoE (~10B active)
Context window
128K
Modality
Text
License
Apache-2.0
  • API
  • Reasoning

State-of-the-art open model for Indian languages, trained on IndiaAI compute

Open weights 2026-02-18 Not disclosed

AI4Bharat

Released

Version1.0

Parameters
3.36B
Context window
N/A
Modality
Text · Audio
License
Apache-2.0
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

TTS for 22 Indian languages with 45 voices

Open weights 2026-09-12 Free / self-hosted

United Kingdom

2 1 organizations

Stability AI

Released

VersionSPAR3D

Parameters
Not disclosed
Context window
N/A
Modality
Image · 3D
License
Stability Community License
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

Turns a single image into a textured, editable 3D mesh in 0.7 seconds

Open weights 2025-01-01 Free / self-hosted

Stability AI

Released

Version3.5(Large / Turbo / Medium)

Parameters
8B / 2.5B
Context window
N/A
Modality
Text · Image
License
Stability Community License
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API

The latest official SD series; SD4 has not been released

Open weights 2024-10-22 Free / self-hosted

Vietnam

2 2 organizations

Viettel AI

Released

Version120B-A12B

Parameters
120B MoE (12B active)
Context window
Long context
Modality
Text
License
—
  • API

Vietnam's sovereign flagship, built on Nemotron 3 Super

Closed source 2026-02-01 Not disclosed

VinAI PhoGPT

Language

VinAI

Released

Version7B5

Parameters
7.5B (also 4B)
Context window
Not disclosed
Modality
Text
License
Llama 2 License
  • API
  • Web
  • Reasoning

The top open generative model for Vietnamese, self-hostable on one card

Open weights Date not found Free / self-hosted

Israel

1 1 organizations

AI21 Labs

Released

Version1.7

Parameters
398B MoE (94B active)
Context window
256K
Modality
Text
License
Apache-2.0
Knowledge cutoff
2024-08-31
  • API
  • Web
  • Reasoning
  • Functions
  • JSON
  • OpenAI API
  • Caching

Hybrid Mamba-Transformer architecture built around long-context text

Open weights 2025-07-01 Not disclosed

Indonesia

1 1 organizations

Sahabat-AI 70B

Language

GoTo / Indosat

Released

Version70B

Parameters
70B (also 8B)
Context window
Not disclosed
Modality
Text
License
Apache-2.0
  • API
  • Reasoning
  • OpenAI API

Open source for Indonesian, already deployed in government service portals

Open weights Date not found Free / self-hosted

Brazil

1 1 organizations

Maritaca AI

Released

Version4

Parameters
Not disclosed
Context window
128K
Modality
Text
License
—
  • API
  • Web
  • Reasoning
  • Functions

Best-in-class for Brazilian Portuguese law practice, leading on the OAB benchmark

Closed source Date not found Not disclosed

Singapore

1 1 organizations

SEA-LION v4.8

Language

AI Singapore

Released

Versionv4.8(120B-A12B)

Parameters
120B MoE (12B active) / 30B-A3B
Context window
Not disclosed
Modality
Text
License
MIT
  • API
  • Functions

Open MoE for 11 Southeast Asian languages, built with NVIDIA

Open weights 2026-09-18 Free / self-hosted

Saudi Arabia

1 1 organizations

SDAIA ALLaM 34B

Language

SDAIA

Released

Version34B

Parameters
34B
Context window
4K
Modality
Text
License
—
  • API

Saudi Arabia's national Arabic model, in production in HUMAIN Chat

Closed source 2025-08-25 Not disclosed

Thailand

1 1 organizations

SCB 10X

Released

VersionS

Parameters
Not disclosed
Context window
Not disclosed
Modality
Text
License
Apache-2.0
  • API

The latest open research preview for Thai

Open weights Date not found Free / self-hosted

Switzerland

1 1 organizations

Apertus 70B

Language

Swiss AI Initiative

Released

Version1.0 70B

Parameters
70B (also 8B)
Context window
Not disclosed
Modality
Text
License
Apache-2.0
  • API

Fully public training data, designed for EU AI Act compliance

Open weights 2025-09-01 Free / self-hosted

Beyond models: agent appsAgent apps

The models above are the fuel; this section lists the vehicles — agent apps of every form (CLI, IDE, plugin, web and platform) and the models behind them. A hand-picked sample, not the whole market — 19 in total.

Anthropic

Built onClaude family

Terminal-based coding agent: reads the repo, edits files, runs tests and opens PRs.

Closed source

OpenAI

Built onGPT family

OpenAI's coding agent, runs in the local terminal or a cloud sandbox.

Closed source

Google DeepMind

Built onGemini family

Google's open-source command-line agent with a 1M-token context and a generous free tier.

Open weights

Qwen Code

CLI

Alibaba

Built onQwen family

Alibaba's CLI coding tool built on Gemini CLI, tuned for Qwen3-Coder.

Open weights

Cursor

IDE

Anysphere

Built onMulti-model (Claude / GPT / Gemini)

The IDE that made agent mode mainstream: Tab completion plus multi-file Composer.

Closed source

Windsurf

IDE

Windsurf

Built onMulti-model

Cascade pipelines tasks step by step and keeps a full trail of agent actions.

Closed source

Trae

IDE

ByteDance

Built onDoubao / Claude

ByteDance's AI IDE: Doubao models in China, Claude available overseas.

Closed source

GitHub · Microsoft

Built onGPT / Claude

The veteran plugin: editor completion, agent mode and code review.

Closed source

Dify

Platform

LangGenius

Built onAny model (platform)

Open-source platform for LLM apps: visual orchestration, RAG and agent workflows, self-hostable.

Open weights

Coze

Platform

ByteDance

Built onDoubao

ByteDance's bot builder: drag-and-drop agents, publish to Lark, WeChat and Discord.

Closed source

n8n

Platform

n8n

Built onMulti-model

Open-source workflow automation: 400+ integrations with LLM nodes for agent steps.

Open weights

OpenHands

Platform

All Hands AI

Built onAny open model

Open-source autonomous agent framework: writes code, browses the web, runs tasks.

Open weights

Manus

Web

Monica

Built onClaude / Qwen

A general-purpose web agent: give it a goal, it plans and executes in a sandbox.

Closed source

Devin

Web

Cognition

Built onIn-house models

The full-time AI software engineer: takes tasks, spins up environments, ships PRs.

Closed source

Aider

CLI

Paul Gauthier

Built onMulti-model (Claude / GPT / DeepSeek)

Terminal-based AI pair programming tool with git-aware multi-file edits.

Open weights

Goose

CLI

Block

Built onMulti-model (Any LLM)

Open-source extensible autonomous agent by Block, natively powered by MCP.

Open weights

Zed

IDE

Zed Industries

Built onClaude / GPT family

High-performance Rust-based editor with built-in multi-model agent workflows.

Open weights

LangGraph

Platform

LangChain

Built onPlatform (Any model)

Framework for building resilient, stateful, multi-agent workflows with cyclic graphs.

Open weights

CrewAI

Platform

CrewAI

Built onPlatform (Any model)

Framework for orchestrating role-playing autonomous AI agents.

Open weights

Models here, skills over there.Now that you know the models, what next?

This page answers “what models exist”; the skill index answers “what ready-made skills can I use”; the authoring guide answers “how do I write one myself.”

Go to the skill index